At 04:44 PM 2/25/2011, Jed Rothwell wrote:
I wrote:
However, I just meant that the margin of error bars are marked at
the bottom of the graph along with the blue line for light water,
and the red heavy water line is far above that margin.
Plus, PLUS! the red line is beautifully correlated with a control
factor, current density. You can see that at a glance. A correlation
is an important way to separate noise from meaningful data.
Yes. Correlation demonstrates probable common cause. However, note: a
correlation between current (or current density, pretty much the same
thing) and power is not surprising! Just from that correlation, in
fact, Kort asserted that the problem was input power estimation error
due to bubbling, which would, indeed, by itself, produce this effect.
If there were enough noise.
I pointed out the relative lack of elevation with hydrogen, and Kort
-- correctly, I think -- pointed out that hydrogen bubbles would be
smaller, because they would be more bouyant (twice as bouyant), and
so there would be less influence caused by them, perhaps, on the
current. While I wasn't convinced -- no thorough analysis was done --
it was, at least, plausible, that deuterium and hydrogen would behave
differently in this respect. I was suspicious, however, that the
difference would be as great as shown in the graphs. Perhaps there
might be half as much bubble noise, half as much input power error,
which would have shown very clearly.
His theory was inconsistent with the time behavior of both cells, and
when we realize that the same current excursion, also under
conditions of high loading, produced a very different result, no
excess heat, the first two times this was run, with hydrogen and
deuterium behaving the same, when we realize that there were platinum
cathode calibrations that would also have shown the bubbling -- with
no loading issue -- the theory came unglued.
Not that this made any difference to Kort! Even though Britz has now
issued a study of the influence of bubble noise on input power
estimation, Kort is still trying to keep this theory's head above
water, saying that he's in communication with Britz about "errors" in
the Britz analysis.
Britz was very thorough. But Britz can be summarized like this.
Bubble noise has little power over 3 KhZ. The power supply
electronics can handle about 1 MHz. The bubble noise has only the
tiniest of effects on the power supply, this source of error can be
disregarded.
In conversations with Kort, I'd stated that I doubted there would be
noise over 10 KHz, based on an understanding of how bubbles would
actually shift resistance, and that the power supply could handle at
least up to 100 KHz. I was being very conservative with
seat-of-the-pants estimates, based on general experience. Kort simply
disregarded all this, and, in fact, made claims that the frequency of
the noise was irrelevant. He did a nice demonstration of a particular
kind of error that Feynman liked to skewer: over-reliance on math. So
I'll examine this a little, for fun.
Kort is correct in that the energy involved in a transition from one
state to another is independent of the time the transition takes.
That's implied by the laws of thermodynamics. Kort justified his
statement by appeal to very complex math that I didn't bother to try
to understand, not that it would have been impossible, just
time-consuming for me, and I didn't doubt the analysis itself, but
how it was being implied.
A transition involves so much energy. Power is energy per unit time.
Energy doesn't depend on the time a transition takes, but power does.
If a transition happens once, and back, in the entire experimental
period, the "AC power" is very, very tiny. If it happens a million
times a second, it could be very large. The *total energy* from AC
power, then, would depend very much on how many transitions there
were in the experimental period.
Kort lost himself in his math, believing that he'd correctly analyzed
the situation, fooling himself, then, with a belief in the solidity
of his theory because it was firmly grounded in "math." And he
assumed that if his math were correct, he must be correct, and he
ridiculed my counterclaims because, he claimed, I was arguing with
"math." And he extended this into a whole universe of claims.
McKubre, he confidently asserted, didn't believe that telephones
worked. Obviously, McKubre believes that telephones work!
What was Barry doing here? He saw -- correctly -- a kind of analogy
between bubble noise and the operation of a carbon microphone. Phones
transmit AC power, based on the resistance variation of the carbon
mike, as the granules are compressed or relaxed by sound pressure.
Kort asserted, again and again, that McKubre didn't believe in "AC
power," nor did I, when, in fact, both McKubre and I specifically
acknowledged AC power.
Kort simply missed that a constant current power supply is immune to
low-freqency resistance noise. I could see his error every step of
the way, but he was so attached to finding a conclusion that made
everyone else into stupid idiots that he couldn't grasp i.t
We fool ourselves. It's a cautionary tale for all of us. I'm pretty
solidly convinced of the reality of cold fusion, by an unknown
reaction that converts deuterium into helium. But if I become
attached to that conclusion, I can easily fool myself.
I'm convinced by Pam's neutron results. But I'm intending
replication, and if I become attached to outcome, I'm screwed. I'm
setting up experimental conditions that should be possible to
*easily* replicate. Because I'm using new material, I can't have a
fixed way of interpreting the results. That's hazardous! That's the
kind of thing that caught Oriani. If you can vary the analysis to fit
the results, you can easily create an impression of solidity where
there isn't anything real. Maybe what I'll see will be so clear that,
obviously, that won't be a cause.
But, then, the real proof will come in replication *of the exact
experiment,* as far as it can be made exact. This time, with the
replications, there will be expected results, and a pre-determined
method of analysis. And if I then sell these kits and others run
them, and these results are published, we will have
multiply-redundant independent findings. Only one small detail would
be left, if anyone wants to dot the i's and cross the t's. My design
and materials will be fully documented. If one wants a completely
independent demonstration, they would obtain the materials and
fabricate the cell themselves. But that might be overkill, for such a
small detail in the overall picture of cold fusion, the production of
neutrons at very low rate with a gold co-dep cathode.
Or I don't find anything! *That will also be interesting.* Don't
worry, folks, I won't jump to conclusions. There are hundreds of ways
to do this wrong, and there is a good chance I'll find at least one
of them, the first time.... Every mistake, though, if investigated,
will build the body of evidence.
Just think of what it would have been like if the negative
replicators in 1989 had not stopped with "negative," but had then,
trusting that the original report wasn't Totally Stupid -- normal
scientific courtesy and prudence -- they had kept trying. There would
have been obvious steps to take. They did not take them, and the
result was 20 years of missed opportunity, for the most part.
