| Topic: |
Science > Physics |
| User: |
"" |
| Date: |
11 Sep 2006 02:50:13 PM |
| Object: |
"Failed" experiments? |
I came across an odd attitude the other day about physics. I run a
high tech company involved in x-ray optics and an employee was decrying
the large number of "failed" experiments as being a waste of financial
resources. This bothered me for days till I realized why. Experiments
are only failures if they are done so they are unable to measure what
was intended, ie, a flawed experiment. A null result is as much a
success as the result you may have been hoping for, sure, its hard to
publish a null result but this is business; a null result tells me what
NOT to do and in almost all cases I have been able to use such a null
result later.
Quite frankly, most of what we do "fails" in the sense that it doesnt'
work in the way we wanted but I expect that and accept it. I am
perfectly willing to put a lot of money into such failures as long as
there is reasonable chance of success and the experiment is well
thought out. I also expect that such "failures" will be useful later.
.
|
|
| User: "" |
|
| Title: Re: "Failed" experiments? |
11 Sep 2006 03:47:23 PM |
|
|
In article <1158004213.277618.41530@b28g2000cwb.googlegroups.com>, writes:
I came across an odd attitude the other day about physics. I run a
high tech company involved in x-ray optics and an employee was decrying
the large number of "failed" experiments as being a waste of financial
resources. This bothered me for days till I realized why. Experiments
are only failures if they are done so they are unable to measure what
was intended, ie, a flawed experiment. A null result is as much a
success as the result you may have been hoping for, sure, its hard to
publish a null result but this is business; a null result tells me what
NOT to do and in almost all cases I have been able to use such a null
result later.
Quite frankly, most of what we do "fails" in the sense that it doesnt'
work in the way we wanted but I expect that and accept it. I am
perfectly willing to put a lot of money into such failures as long as
there is reasonable chance of success and the experiment is well
thought out. I also expect that such "failures" will be useful later.
A little story about Edison comes to mind. During the press
conference, where he announced the success of the light bulb project,
he mentioned that his people tried 7000 different substances till they
found one good enough for a filament. So, one reporter asked "do you
mean to say that you failed 7000 times?". "No", responded Edison, "I
mean to say that we succeeded to learn that 7000 potential candidates
do not work." Couldn't think of a better answer.
Yes, something failing to work is a data point, and a valuable one.
Knowing what doesn't work is just as important as knowing what does.
Unfortunately, the way science publishing is structured, it
discourages the reporting of "failures", and that's why enormous
efforts are wasted on "reinventing wheels" and repeating other's
errors. And, that's why, while one can become a theorist by studying
from textbooks and journal articles, experimentalists train the "old
fashioned" way, by apprenticing to an experienced "master". Journals
and books may teach you what to do, but you need the "master" to teach
you what not to do.
Mati Meron | "When you argue with a fool,
meron@cars.uchicago.edu | chances are he is doing just the same"
.
|
|
|
|
|
Related Articles |
|
|
