James Hamlin talks about Data Issues in High-Pressure Superconductivity Papers
Vložit
- čas přidán 7. 03. 2023
- This talk was recorded at the online workshop 'Does Condensed Matter Physics need to worry about a reproducibility crisis?', on March 2nd 2023.
virtualscienceforum.org/repro...
Can we please have a round of applause for the effort Prof. James Hamlin made for seeking the truth behind the claims? He has done a phenomenal scientific investigation, and clearly voiced out both his educated guesses and outright facts from the publications in consideration. This is what good science is all about.
We all know this talk is pointing directly at Ranga Dias. LOL 😁
This is a remarkable talk and I enjoyed it even though I am not in this field. Regardless I can’t help that the part I’ll remember most is the last 15 seconds 😂
High profile journals, Nature, Science etc should be routinely employing the services of statisticians to screen for malpractice in data handling. Imagine the time and effort of other groups wasted when trusting the results of such fraudulent work. In fact I met someone at a conference who said he spent most of his PhD on the basis of the work of H. Schon.
The trend might be to publish your attempt to resolve a question with the way you are handling the data, and later on you are not allowed to deviate from what was published. So you don't change your analysis routine after you've seen some kind of result. The result you get is the one you publish (or don't).
Thank you for posting this interesting video!
Interesting. The same group just published it again.
As a paratrooper in this field I wish to thank you for the pdf2svg :D
Gosh. Thanks for providing this.
As a regular guy who knows nothing about this, Why did they public fake data then? To be famous? Is this even legal? Will science community punish them?
I really don't understand, Why are they doing this?
Those of us who grew up loving science and admiring scientists tend to think that researchers are some breed of sages, who make back-breaking effort all their lives in exchange for dirt cheap money. Turns out, scientists are just as susceptible to avarice as the rest of the human population.
The harshest punishment that scientific community offers is that of irrelevance. Essentially, once you're found out as a fraud, no one would be willing to collaborate with you, or corroborate and build up on your work. Essentially, you'd be outcasted.
@@sjhd98 Unfortunately there are many non-scientistists with money willing to believe and reward such behavior.
@@gilbertanderson3456 Yes, I agree with you. We're living in those times where the "perception" of scientific progress is more important than the "actual" on-ground scientific progress. Artificial Intelligence is one such area. Most non-scientists believe we have invented God by building a large-scale token predictor, which goes by the name of Large Language Model.
I always publish all python script and raw data associated with a paper in a separate DOI minted repository. It's not that hard...
Ranga Dias would disagree
Same group just published another paper in Nature.
The pressure to publish, and the pressure to make graphs "look good enough" to be published can push research groups to do things thar are far from ideal in terms of best practice in physics. This is not surprising at all, and we are talking about pretty big groups where the data passes through many people before it ends in the paper. The recent result at 1GPa will rightfully be put through scrutiny, but to dismiss it outright based on past mishaps seems extreme too, to me.
what about comments to this blogpost? nanoscale.blogspot.com/2023/03/aps-march-meeting-2023-day-2.html
@@spinespresso I read through the comments and yesterday I thoroughly read the whole paper including all the supplementary, and the past two years I have been reading papers on the topic of superhydride superconductivity as they come out.
About the comments, first lets filter out some silly "issues" like why data doesn't go down to 0K. It doesn't cos measuring to 0K is hard, and temperature drop down to what your device can do or so. However looking at the comments in aggregate there seems to be some substance to the alarming issues, especially in regards to past misconduct on MnS2 and so on. As I said this does not directly reflect on the current topic but it's relevant.
Most illuminating is the least hyperbole suggestion- what if this is not a superconducting but a diamagnetic metal/semiconductor transition, that may be in line with LuH2/3 type situations seen elsewhere? Of note- the measurement ls were Not done under H2/N2 atmosphere, only fabrication. Doesn't change the possibility but the commenter got That part wrong.
Overall the claims of Ranga et.al, from the pov of someone reading the stream of papers, are out of left field. A result like LuBeH8 (or somesuch in a similar vain) would line up cleanly with theoretical strides of the past 10 months. LH3 (L=lanthanide) analogue is an extraordinary happening- the parent material does not have the H atoms form any motif linked to pushing the H electrons to the fermi levels.
The paper does attempt to support with extraordinary level of detail the extraordinary result. But looking at the human element with a company's future riding on this result, interpretation of the data us overwhelmingly biased towards "Superconductivity!" Even if no misconduct is to be had, but with another physical transformation causing the result.
I have no verdict. How can I have one? It's been a day since the publication. This is exciting and scary at once with the information I have. I will wait patiently, read and dismiss the initial drama that newstellers will spew in the following weeks, and we will get answers in about 6 months haha.
@@mishael1339 thanks for your comments. do you mean the cell cannot cool below 100K?
@@spinespresso I do not know for sure. If I recall correctly they cooled with Thermoelectric cooling (paltier device), which (referencing Wikipedia) creates waste heat itself according to ohm's law and typically cools 70 degrees Celsius for a 1-stage cooler. Reasonably this isn't a system that's capable of reaching ultra low temperatures, and if they used Liquid Nitrogen the lowest temperature would be 70K or so as well. They did many measurements during the reheating which is the more stable operation side. So yeah, it's completely reasonable that they used simpler setups that don't go below 100K, using a liquid helium criostat is a needless hastle if that's not where your transition is. I don't see the conspiracy here, personally.
@@mishael1339 ok thanks. it should be possible to find out, or ask, which exact setups were used. and also ask to remeasure in PPMS if that is not the one that has been used.