The disappearance of darkness

J. Clarke wrote:
In article , ozcvgtt02
J. Clarke wrote:

Would you be kind enough to provide an example of an experiment in which
"assumptions and models" affect the outcome?

Experiment as in "the measured raw results" or as in "the
results after evaluating the measurements"?

As in the published paper.

Almost all data that was used to conclude "facts" that were
later found to be not that way.

-Wolfgang

In article , ozcvgtt02
@sneakemail.com says...

J. Clarke wrote:
In article ,
In rec.photo.digital.slr-systems J. Clarke wrote:
In article ,
In rec.photo.digital J. Clarke wrote:
In article , ozcvgtt02
J. Clarke wrote:

Would you be kind enough to provide an example of an experiment in which
"assumptions and models" affect the outcome?

Experiment as in "the measured raw results" or as in "the
results after evaluating the measurements"?

As in the published paper.

Do those published papers count in which the authors later revised the
conclusions they originally drew from their experimental results? Or
those in which later reviewers, not the original authors, did the same
thing?

Give us your example if you have one.

The famous example is the Millikan-Ehrenhaft controversy, e.g.:

http://www.umich.edu/~chemstu/conten...n_Erenhaft.pdf

http://bjps.oxfordjournals.org/content/56/4/681.short

Both the above have many good further references.

And the upshot of that is that Millikan fudged the data,

BECAUSE he had "assumptions and models" he wanted to appease.
The measured raw results were ... interpreted and filtered in
the published paper.

They were not "interpreted and filtered". Data points that did not
support his viewpoint were rejected. That is not experimental science,
that is religion. He's an example of a good theoretician who was not a
good experimentalist.

which is a nono
in any experiment, regardless of the "assumptions and models".

So, according to our current knowledge, is charge quantized
(Millikan, fudged) or not (Ehrenhaft, not fudged)?

According to the current not fudged data, replicated many, many, many
times, it is quantized.

J. Clarke wrote:
In article , ozcvgtt02
J. Clarke wrote:

And the upshot of that is that Millikan fudged the data,

BECAUSE he had "assumptions and models" he wanted to appease.
The measured raw results were ... interpreted and filtered in
the published paper.

They were not "interpreted and filtered". Data points that did not
support his viewpoint were rejected.

What do you think a filter/filtering something does?


That is not experimental science,
that is religion. He's an example of a good theoretician who was not a
good experimentalist.

Religion is being a good theoretician?


You know the old joke?
A quack doctor (in the then Wild West) travelling from
town to town treated a farmer for fever: "Eat sauerkraut".
The farmer recovers. The quack doctor writes in his little
black book: "Sauerkraut helps against fever."

Another town, soon after. The smith has fever. So,
consulting his little black book, he orders him to eat
sauerkraut. The smith dies, however.

The quack doctor corrects the entry in his little black
book: "Sauerkraut helps against fever only for farmers."

Can you spot the model and assumption that led the quack doctor
to the "corrected" entry?


which is a nono
in any experiment, regardless of the "assumptions and models".

So, according to our current knowledge, is charge quantized
(Millikan, fudged) or not (Ehrenhaft, not fudged)?

According to the current not fudged data, replicated many, many, many
times, it is quantized.

So basically Ehrenhaft was wrong.

-Wolfgang

In article , ozcvgtt02
@sneakemail.com says...

J. Clarke wrote:
In article , ozcvgtt02
J. Clarke wrote:

And the upshot of that is that Millikan fudged the data,

BECAUSE he had "assumptions and models" he wanted to appease.
The measured raw results were ... interpreted and filtered in
the published paper.

They were not "interpreted and filtered". Data points that did not
support his viewpoint were rejected.

What do you think a filter/filtering something does?


That is not experimental science,
that is religion. He's an example of a good theoretician who was not a
good experimentalist.

Religion is being a good theoretician?

No, picking your data to support your model is behaving like a religious
zealot.

You know the old joke?
A quack doctor (in the then Wild West) travelling from
town to town treated a farmer for fever: "Eat sauerkraut".
The farmer recovers. The quack doctor writes in his little
black book: "Sauerkraut helps against fever."

Another town, soon after. The smith has fever. So,
consulting his little black book, he orders him to eat
sauerkraut. The smith dies, however.

The quack doctor corrects the entry in his little black
book: "Sauerkraut helps against fever only for farmers."

Can you spot the model and assumption that led the quack doctor
to the "corrected" entry?

As an experimentalist I'm interested only in the data, not the model.

which is a nono
in any experiment, regardless of the "assumptions and models".

So, according to our current knowledge, is charge quantized
(Millikan, fudged) or not (Ehrenhaft, not fudged)?

According to the current not fudged data, replicated many, many, many
times, it is quantized.

So basically Ehrenhaft was wrong.

His theory was wrong, his experiments though were simply imprecise.

Science doesn't work by isolated experiment. We had two experimenters,
one of whom fudged the data and the other of whom was sloppy, so they
got different results. Later experimenters did not fudge the data and
were not sloppy and found that their results were similar to those
obtained by the one who did fudge. At that point his model was
confirmed. If instead of fudging his data he fixed his experiment then
he would not have had to fudge.

J. Clarke wrote:
In article , ozcvgtt02
J. Clarke wrote:
In article , ozcvgtt02
J. Clarke wrote:

They were not "interpreted and filtered". Data points that did not
support his viewpoint were rejected.

What do you think a filter/filtering something does?

That is not experimental science,
that is religion. He's an example of a good theoretician who was not a
good experimentalist.

Religion is being a good theoretician?

No, picking your data to support your model is behaving like a religious
zealot.

I'd offer that many *other* types or zealots do that.
Politicans do, even a-religious, a-moral ones.
Most people in a job interview select the parts to tell and
to be silent about --- even when they say no untrue word and
colour nothing --- trying to give a most positive impression
of themselves. Are they relogious zealots, too?


You know the old joke?
A quack doctor (in the then Wild West) travelling from
town to town treated a farmer for fever: "Eat sauerkraut".
The farmer recovers. The quack doctor writes in his little
black book: "Sauerkraut helps against fever."

Another town, soon after. The smith has fever. So,
consulting his little black book, he orders him to eat
sauerkraut. The smith dies, however.

The quack doctor corrects the entry in his little black
book: "Sauerkraut helps against fever only for farmers."

Can you spot the model and assumption that led the quack doctor
to the "corrected" entry?

As an experimentalist I'm interested only in the data, not the model.

The data clearly says that Sauerkraut only helps farmers with
fever! You *need* a model saying that usually the occupation
(or to which gods they pray or if they wear their hair long or
short) does not influence the illness, but that other factors
(say, germs or impact trauma) do.

Unless you have a model, you can't even check if the data
supports it or create a test to check your model.

which is a nono
in any experiment, regardless of the "assumptions and models".

So, according to our current knowledge, is charge quantized
(Millikan, fudged) or not (Ehrenhaft, not fudged)?

According to the current not fudged data, replicated many, many, many
times, it is quantized.

So basically Ehrenhaft was wrong.

His theory was wrong, his experiments though were simply imprecise.

Obviously Ehrenhaft (I still don't get over his name! Translate
it to English one day ...) *thought* his data was precise
enough. Either he was a bad experimenter, or his experimental
apparature or process was faulty, or he didn't drop data that
was clearly invalid (which is, by some definition, fudging,
yet not doing so is also clearly fudging).


Science doesn't work by isolated experiment. We had two experimenters,
one of whom fudged the data and the other of whom was sloppy, so they
got different results. Later experimenters did not fudge the data and
were not sloppy and found that their results were similar to those
obtained by the one who did fudge. At that point his model was
confirmed. If instead of fudging his data he fixed his experiment then
he would not have had to fudge.

In fact, it was a tiny bit more complicated than that.
Following experimenters *also* fudged their data in sort of
the same way Millikan did: Millikan gave too low a charge for
the electron, follwing experimenters reported a slighly larger
number. Then following experimenters *also* reported a still
slightly larger number. THEN follwing experimenters reported
yet STILL slightly larger numbers. And so on. (If they hadn't
fudged, you'd not find an asymptotically approach to today's
number, but values straddling the final number ...)

Today the number has stabilized; it's more than 5 times larger
than Millikan's standard error on his data.

-Wolfgang