How to read LENR experimental results

Eric,

You, like others here, are concentrating on the criticism. You are asking: "can a valid flaw in this experiment be proved". (Your argument is actually - look it is easy to make unsound criticism, so how can I know your criticism is sound?).

My perspective is different. I'm sure you will understand this if you've ever conducted experiments on systems where the results are predicted by theory but the theory uses idealisations and there are always other effects that require more sophisticated use of theory with parameters not easy to determine.

The errors that actually happen are often surprising, and not easy to predict. All you can do is tighten up every aspect of the experiment that could possibly introduce an error - and for that you need a pretty sophisticated model of the errors.

The question I, and 95% of scientists looking at these matters, want to ask is:
"Can it be proved that the results are beyond possible reasonable error?"

This is a much stronger requirement.

I'm pretty sure the difference here is that i think this is appropriate, and many others (possibly you) do not.

I realise you have your reasons for this view - let me just review a few I can think of, and why they do not sway me:

(1) The LENR phenomena is a diverse set of anomalies that individually might not be convincing but collectively are.

If these anomalies had coherence this would be a strong argument. Data mining can discover important patterns that are not significant in individual cases. However there is no coherence about these anomalies except for the sign of the excess heat. That is 50% likely correct and you will appreciate that any LENR apparatus that showed persistent negative power out will be (correctly) recalibrated and not presented as an LENR research paper. The only coherence about these anomalies is that all close to experimental possible error - that surprisingly remains true for both inaccurate and accurate calorimetry - the opposite of what would be expected were the anomalies real.

(2) Even though these anomalies are not coherent it is quite unlikely that any single one should be observed, the likelihood of the ensemble is thus the product of these and therefore very unlikely.

The issue here is about independence and selection. If each result was giving independent unbiassed information this argument would run. In reality the results are selected and dependent in the sense that:
(a) LENR techniques that generate positive results tend to propagate and be replicated, understandably. But this applies equally to error artifacts as to LENR anomalies. (Known artifacts will not so easily propagate. But methods like variants of isoperibolic calorimetry where artifacts are difficult to avoid but also difficult to prove will flourish).
(b) If this were true the experiments with better calorimetry would show similar and therefore more significant excesses, in reality they show smaller excesses.
(c) Both at the level of reported results, remembered results, and individual experiments, it is natural for positives to be selected. For example, if a researcher studying these Zr/Ni alloys performs an experiment that shows no excess heat (as they did here with lower temperature tests) they vary some parameter (in this case temperature) until they get the anticipated excess heat. You only need methodology which gives errors under some conditions, and those conditions can be found by a natural experimental process of optimisation. These process happens both individually, and collectively, as "successful" experiments are copied. the analogy with selection of the fittest is strong and accurate.

(3) I'm going to draw out of this one factor, where I make a prediction. I predict that for replicable experiments (there are quite a number now) the excess heat found will vary with the methodology, thus "good" mass flow calorimetry will show smaller excess heat. I define excess hereas ratio of excess power to power in over a long test period, since tests over shorter period have many issues.

There may of course be a few claims of mass flow calorimetry positives that are just obviously wrong - for example the guy - who was it? - who claimed an excess that correlated exactly with the power lost from the pump heat dissipation - because the pump housing was thermally connected to the system and the pump was very inefficient so that the electrical power dissipated was many times the mechanical power added. The mistake was to consider mechanical power added not electrical power dissipated.

Quote

Thomas, there is no point in further debate If we cant even agree om the basic basic science of heat exchange.You did not reply to my comment, because you agree may be?"Thomas, your last post surprises me, did you think it through ?If you have a box, pipe or whatever container and measure the outer wall temperature, the heat flow is governed by the temperature difference between outer wall temperature and surroundings by conduction, convection and radiation to the surroundings.It does not matter how the internals look like, number of walls, chambers, heaters etc.An even outer wall temperature will therefore have a certain heat flow to the surroundings by conduction,convection and radiation. Therefore a calibration curve will work of outer wall temperature vs internal power.And therefore it does not matter If the heat arrives to outer wall from the electrical heater or from the inner core as LENR heat.

So - a brief reply.

(1) Yes, based on your post here I have a more detailed understanding of how to determine heat flows in this system from observed temperature measurements. But, no, I have not thought it through. That takes a long time. It is actually why I enjoy posting here - I like understanding, in detail, new things. I do find that even when I know I'm not fully understanding them I do better than most people. In case that sounds arrogant I also find when I understand things better I can be embarrased at how long that took me.

I guess it is because I tend to understand stuff from first principles: so I know until I've done this the concepts I'm using can be abused. I'd accede to your superior understanding except I've as yet seen no evidence of it - details can be quite subtle and need more than words to manage.

(2) Your last comment. I guess i did not see it, or I just got distracted by other tangential stuff on this thread. This sub-thread - though I'm not coping out of it (no pun intended) is actually a distraction from the original matter which was your Cu/Ni/ZrO2 extraodinary temperature anomaly claims. I'm aware that I promised a decent analysis of that (I have it in my head just not written down) also a decent analysis of Fleischmann's paper and the critique (I have this 50% in my head).

(3) I'm posting here for fun so which of these projects gets done sort of depends on my whim but if otehrs feel there is some crucial killer thing that is most important that would be enough to sway me one way or another.

(4) Although I like details (which I'm good at simply through persistence and an ability to cope with any maths), I also like the subtle use of language and consideration of probabilities epitomised here by Eric. So I'll probably continue for a while refining the differences between me and Eric over how we evaluate LENR plausibility.

(5) "No point on further debate". I'm not sure why you say that. I guess it is because you think I have nothing to contribute here because your understanding is so good. That may be true but you'd need to write it down clearly in an detail - that would save me the job of doing the same. Personally, I think it is the height of arrogance to refuse in principle further debate except where the person wanting this is clearly offering no substance. Understanding can always be improved.

(6) Your contribution to this debate could be to settle on one paper. I don't mind reading references - in fact with the topic of this thread I looked at [6],[7] before you replied - your posting the "real" [7] was helpful - though that reference adds less than you might think to the paper itself. If we focus on one paper we can really get to the bottom of what it means. My default is to continue with the topic at the top of this thread since that is not exhausted for example:
(1) what are the calorimetry error bounds (in W/g Ni and in eV/atom Ni) for the apparatus. How were they calculated? (I know but can't accept the easy answer).
(2) what is the maximum chemical energy from this system? What assumptions does this maximum require?
(3) what is the meaning of the dynamic measurements? I've stated what I think it is, you've said I'm wrong based on the references but not put forward anything different. I think it would be courteous for you to do that.

I realise that it is easy to take a broad brush look and say that the precise details are irrelevant. That may be so, but I find it is investigating those precise details that unlocks further insight into what goes on here. I therefore won't accede to your broad brush comments being better than mine unless they are backed by details (in some cases I've done this myself for my own comments, but there is more to do and as you say I have to read carefully a few of the refs not just skim them).

Quote

It was the annealing part that inspired the alien analogy (Although might be a little unfair when discussing in terms of "5 hour power spikes").

I was looking generally for state changes with hysteresis modulated by temperature. Annealing qualifies.

Now if you want a specific mechanism (one of many)...

Annealing can alter surface emissivity that significantly changes radiative heat transfer. Is this relevant? Probably not. Can you prove it? Well, it would take you a very long time...

My gut feeling here is that there is some hysteretic change here we are just not thinking of - which is why I went to general classes of mechanism.

Quote

@Tom,Concerning isoperibolic calorimetry as a general topic, beyond the paper by Focardi et al. referenced above, you should bring your analysis of its inherent problems to the Vortex list and defend your thesis there. (This is a trap: you will be an innocent lamb waiting for Jed Rothwell to pull it apart into tiny pieces and methodically deconstruct each one. He is very knowledgeable on the topic. He may even agree with you on a point or two.)Eric(Curious enough about the topic, I've gone ahead and submitted a question of my own, with a link to your claims.)

I don't doubt Jed is more knowledgeable than me. I tend to avoid vortex, because it is a bit weird and also uncomfortable technologically to use. I'm not claiming isoperibolic C is always unreliable just that as used here it is unsafe. Temperature is only a good proxy for heat flow when thermal resistance is well controlled. Things get more complex when the heat flow graph is not linear.

But maybe I should engage with Jed a bit on the experimental stuff. I've never seen myself as an innocent lamb. Perhaps a Socratic figure, since the things I enjoy doing tend to be things I am working out as I go.

Eric

Quote

My point is more subtle. It's that one can come up with interesting ways of discounting a hypothesis quite easily (the criticisms above, about one way to extract core samples, were basically free association). If one spends more time at it, one can even come up with something that seems plausible to oneself.

Here's where things get interesting. Not everyone's opinion is equal. My opinion about what constitutes potential flaws in the calorimetry of a specific experiment (Focardi's, above) or a whole kind of calorimetry (isoperibolic calorimetry) does not hold a candle to the opinions of a smart scientist who has been wrestling with the problem for years.

That does not mean I will always be incorrect when I think I might have spotted a problem. It simply is a statement about probabilities. The person with a lot of firsthand experience is more likely to know what a sound or unsound approach entails. My opinions in the matter are my opinions. I can raise as many subtle points as I want, but unless someone is an expert in calorimetry, people should listen first and foremost to those who are and take my astute remarks with a huge grain of salt, for I am not speaking from a position of knowledge.

Needless to say, it would be ambitious of me to attempt to discount an entire method of extracting core samples whose science is not considered to be controversial by practitioners in the field.

Display More

I can see that my summary of your point was somewhat approximate. However, you are not addressing my point, which is the context in which I made that summary. In that context your fuller statement here applies as well as my approximate summary.

Specifically, while I agree that experts will have better understanding of the known unknowns they may, if engaged in this matter from a particular perspective, not have good judgement over the unknown unknowns. These come into play when considering the difference between "Can't prove there is an error" and "prove there is no error" (you'll forgive me for this slightly innacurate summary of the juxtaposition). Experts in fact are more likely to dismiss the possibility of unusual artifacts unless they are experts in looking for unusual artifacts. Which Jed may be, the question is how much.

I've done some googling. Here is what I think is an excellent back and forth between Josh Cude and Jed Rothwell ocer the status of LENR evidence. It is long, with repeated challenge, response, and counter-response.
The link is here, ignore the article and read the long list of comments. It gets most interesting when Joshua and Jed discuss the challenge of "recent peer-reviewed literature proving LENR".

A (skeptical) bystander summarises this discussion and I'd agree:

Quote

What a wonderful capsule summary of the past two years of internet wrangling over LENR!
Nothing summarizes the nature of the debate better than Joshua Cude patiently deconstructing the unsubstantiated claims of the faithful while Jed Rothwell resorts to calling anyone a fanatic if they don’t surrender to his arguments by quantity over quality.

As was pointed out, cold fusion is not going to get any more real by having more people believe in it or be fans of it. Conversely, it is not going to get any less real no matter how many people “dislike it”. Physical phenomena do not respond to popular demand. If cold fusion is real, then it has to actually work. And even if Jed has 10,000 documents in his database, at least one of them has to be good enough to prove it.

Based on the meta-information here I'd be very confident locking horns with Jed over the calorimetry stuff. i don't think I'm more expert than him, but I'm pretty sure I've got better background knowledge and more ability to go deeply into specifics.

Jed - i don't suppose you'd like to come here to do that? So much more comfortable than vortex...

Another interesting factoid from that thread (2013) is how many people were saying then "well - it is make or break for Rossi - we will know within a few months whether he has the goods or nothing".

Plus ca change, plus c'est la meme chose (sorry no accents).

@Paradigmnoia

Quote

Where are all the negative COP results? If it is mostly errors causing spurious excess heat results, where are the lossy errors? It always seems to be excess heat, not messy heat.

There is a paradigm at work here. If you are looking for LENR you don't spend weeks with experiments that show negative excess heat - you recalibrate your equipment, or investigate errors further till you work out why it is negative. Similarly, you prefer to use types of experiment that maximise your "effect".

But, if you look beyond headlines at the actual raw data you'll see that it is nearly always very messy - as expected from these results in error/artifact territory. Take, for example, the data from the paper that headlines this thread.

@Eric Walker

There are two levels at which the debate on LENR can be carried out - both interest me.

(1) The detailed level - where individual reports are scrutinised and conclusions drawn. For me the most fun part here is doing this myself, but it is also possible to read other people's critiques. Where not skilled, and not willing to put in the work necessary to become skilled, the merits of other people's arguments can best be judged by scrutinising such a critique and counter-critique. You get a good idea of which points are one-sided from what is NOT answered. An example would be the Fleischmann vs Morrison exchanges.

I should point out that I don't claim particular cleverness, or particular skill, in my efforts at this level. However, I do claim good background knowledge, good persistence, the willingness to read with an open mind what other people have said, and an ability to perform critical appraisal of such work. As a result my understanding, both qualitative and quantitative, of the Lugano test results is very good. Most of the issues I consider were first suggested by somone else. My contribution was to feret them out, tie up the details, and see how these fit the evidence. For example HG Branzell's idea that the X3 discrepancy between active and dummy tests was due to erroneous input power measurement from clamp reversal is compelling, and I accepted it as best fit for quite some time, but it did not make me happy. That was because although best fit it did not properly explain details of the data (like the "acceleration") and though these could be dismissed I was not happy to do this. Then other major issue - emissivity - suggested by Bob Higgins and GSVIT (refs [5] and [7] in my paper) fascinated me because neither of these had properly quantified it and I found working out how to do this soundly difficult.

Bob had done quite a bit of detailed analytical work but his analysis was clearly incomplete - the critical step was not contained in his work which jumped from one place to another without the necessary logical connection, and I distrusted what was not proven. GSVIT's analysis was qualitative.

It took me a long time to puzzle this out exactly because I was not an expert - I knew nothing about thermography when I started and my knowledge of statistical mechanics (the Planck curve etc) was very rusty.
What motivated me to write up my ideas properly was the fact that, counter-intuitively, the emissivity issue when properly worked out accounted exactly for the "acceleration" in the data.
That involved no cherry-picking of assumptions - in fact the acceleration phenomena is independent of most of the parameters that move the overall COP up or down.

I've gone into this in detail because it shows how predictivity, or its weaker form, explanatory skill, is crucial in weighing up different hypotheses. Explanatory skill is weaker because it is easy to make hypotheses fit existing facts by cherry-picking assumptions. In this case, although the emissivity hypothesis was not technically predictive, I knew from how i'd worked it out that it was actually predictive. That is because there were no cherry-picked assumptions: the result come out of the numerical calculations ab initio. I then tested this and found the result robust to changes in all my assumptions.

The big problem for LENR, as a hypothesis, is this lack of predictivity. "Experimental error plus artifact" is intrinsically not very predictive, but also intrinsically much more likely. Interestingly - and this relates to point (2) below there is some prediction that "Error +artifact makes" that can be, and is being, tested. If "error + artifact" is true (I'm considering, for simplicity, just excess heat) we expect:
(a) good, robust, repeatable excess heat anomaly results will be of an order that depends on the quality calorimetry used. More robust calorimetry will lead to lower magnitude results. there will be some outliers due to unrecognised artifacts, but these will not be repeatable, even stochastically.
The null hypothesis "not error + artifact" which of course includes LENR but is more general because it does not imply nuclear reactions, makes a different prediction:
(b) good, robust, repeatable excess heat anomaly results will be of an order that has no correlation with the quality calorimetry used. More robust calorimetry will not lead to lower magnitude results. There will be some failures due to lack of control of the necessary conditions for the phenomena to happen, but a specific experiment, with specific conditions, will show stochastically significant results which do not reduce in magnitude as experimental conditions are tightened up.

So: these are the two specific criteria that, independent of other a priori judgements, differentiate "LENR is real" from "LENR is anomaly". I respect those LENR experimenters who (whether they use this language or not) show that they see this to be important.

(2) The meta-level. Here the issue is in how to evaluate the significance of a collection of results and the reaction of many other people, some experts, some not, to this data. This was the topic on which Joshua and Jed exchanged views.

The predictivity of hypotheses, as discussed above under (1) affects this meta-discussion. That was not much discussed by Jed and Josh but a related matter was.

If LENR - in the "nuclear reactions" form is real then it predicts that excess heat will be found in quantities that vary greatly from one experiment to another. this is mandated by the OOM difference between nuclear energies and chemical energies. There is just no way that the available excess heat could always magically scale to be exactly within the range of possible chemical + experimental error because the predicted variation in excess heat over different fuels etc is too large. This is a subtle argument and one that Joshua articulates with great force. It is one that I naturally understand because, as above, I'm interested in the predictivity of hypotheses. You need to have an understanding of how large and small numbers work in practice that is apparently trivial but actually lacking in many people to see the full force of this, and its generality.

Finally, the matter most debated by Josh and Jed was to do with what I'd call "progression". The key differentiator between scientific development and pseudo-science is that the overall results claimed from pseudo-science don't get sharper and more definite with time. Whereas the overall results with scientific phenomena do. Repeated experiments will naturally find better ways to expose the phenomena and measure it - these measurements will more clearly differentiate from known theory. So, for example, issues about the perihelion of mercury were an experimental anomaly that emerged from error because the same correlated anomalies were detected by different groups with different instruments. As time progressed the magnitude and details of the anomaly became clearer, until it was absolutely beyond experimental error. Corrections from other planets became tighter, and it became definite that something in addition to `Newton's theory applied to the solar system was needed to explain it. The specific correction needed was known, so that any candidate theory could be tested quantitatively against that. Clear progression.

Quote

I would not give much weight to the opinion of a physicist with no knowledge of programming or databases concerning whether to use a B-tree or hash index for indexing a compound key on a database. I would give a lot of weight to the opinion of a database admin with several years of production experience, even if they had no college education, and would seek out that opinion if it was pertinent

To extend this analogy, perhaps too far: I'd respect the database admin for practical knowledge of how existing technology applies to a wide range of typical problems. On the other hand I'd respect a Computer Scientist with a good grasp of algorithms and their asymptotic behaviour to evaluate how, comparatively, different commercial algorithms would behave in some new domain different from anything currently used.

Both skill sets have their place, in general both are needed, and of course both may be found in the same individual since some but not all database administrators will have a good algorithms background and the flexibility to see how it applies to different problems.