Detectors like this work on exposure. They're always on (except for calibration and maintenance), waiting for events to happen. This paper was written with 2.8 tonne-years of data. That is, 4.7 tonnes of liquid xenon for a little mmore than half a year. The detector has 7 tonnes, and the 4.7 number reflects cuts they made on parts of the detector that either they don't understand as well, or have higher backgrounds.
As they better understand the detector, they can use more of that mass. They have data from it, but they just didn't use it. And they're always collecting more data, too, as time passes.
So the 3x is saying they have something like 8.5 tonne-years of data.
Hm, they have 7 tonnes of Xenon. Events detected all around in the matter, but the PMTs can localise where the event happened. So they can virtually segment parts of the detector where they are sure all the outside effects are understood and taken care of.
To start with, they do a lot of work to eliminate radioactive backgrounds in the materials they're using, and they put the detector deep underground to shield from cosmic rays.
Additionally, when a particle interacts with the nucleus, the ratio of how much energy ends up as scintillation light versus ionization is different than when a particle interacts with an electron, which is most of the background processes.
Then, whatever is left, they try to model using known processes. After all that, there's one event that they can't account for. And that's what the news is about.
I agree. Moreover, I'm not sure if it's the same team, but in a similar experiment while removing all the other effects, they discovered that Xenon 124 is radioactive, but the half life is super long and no one had seen it before. https://xenonexperiment.org/observing-the-rarest-decay-proce...
Makes me wonder if all atoms with 2+ nucleus elements (protons and neutrons) are radioactive but the halflife is so far out as to make something we'll never detect.
Probably not, they have a pretty good handle on why atoms decay, to the point they can predict some to be radioactive before it's actually observed to be (like Bismuth-209).
Also even if something is REALLY REALLY long lasting, you can still check for the halflife by observing enough of it, they've been able to rule out proton halflives under 10^34 years (the universe is on the order of 10^10 years old) but by observing enough protons (like say 50,000 tons of water) you would expect at least some to decay.
> they have a pretty good handle on why atoms decay
Oh, they actually don't. Radioactive decay, AFAIK, is still an open physics mystery. We know it happens, we don't know why, what causes it, or if there even is a cause. We can predict factors that make it more likely.
> to the point they can predict some to be radioactive before it's actually observed to be (like Bismuth-209).
Right, but Xenon 124 wasn't predicted to be radioactive which is what makes it fascinating. It shows holes in what we can predict as being radioactive which is what makes me wonder about everything being radioactive but the timetable is too far out.
I don't actually think Xenon-124 being radioactive was a surprise. All the publicity related to observing the decay for the first time is phrased around 'hey we observed something that's very rare' not 'hey this thing happened we didn't expect to happen'.
Which seems to point even more towards, scientists have a pretty good handle on which ones are radioactive.
Where is the mystery? Any system can spontaneously transform into a new state with a probability greater than zero unless some conservation law prevents it. In a sense it's just quantum tunneling.
Can you be more specific? I believe quantum mechanics explains all of radioactive decay. Unless the GP meant that QM is mysterious, I don't understand the problem.
Quantum Chromodynamics, path integrals, and what other mechanisms you need for deriving the half life are extremely complicated, especially for that many particles, so of course it's hopeless without approximations, which can be wrong, or are even wrong by definition. But the fact that it's prohibitively hard to solve the equations for such complicated systems doesn't indicate a gap in our understanding. You don't even have to go to xenon, this is already the case for tritium.
Some theories predict that protons decay, but the half life is like 1E31 or 1E35 years (compare to the Xe124 that has a half life of only 1E24 years). All experiments so far to measure the proton decay have failed, anyway. https://en.wikipedia.org/wiki/Proton_decay
I don't remember anything specific about deuterium, and the method that Xe124 uses is not available, and I can't imagine a razonable alternative method, so my guess is that deuterium is as stable as protons.
Yeah, it's funny, for experiments like this you spend 90% of your time modeling and subtracting noise, and 10% analyzing the signal that results. Had the same experience in X-ray astronomy. 3 years building a detailed model of all the sources of noise, then subtracting it out and finally starting on the science.
I worked a little on the Virgo interferometer, I would say about 99.9% of the work on those types of detectors is limiting and subtracting noise. ( ≧ᗜ≦)
I read their preprint[1] and they did a thorough job. They investigated a number of the things I'd suspect if I were looking for mis-reconstructed events or weird backgrounds.
So it's certainly interesting!
That said, particle physics history is full of 3 sigma particle "discoveries" that disappeared with more data. They're collecting more, so hopefully we'll learn more in a few more years.
This sort of thing is really useful for helping people to understand what the purpose of scientific publication is. It's not about presenting finished products to society, it's more like a Discord where you communicate new findings with other authors. Often the reason for a communication is because you found something weird, and you want other people to know about it so they can help you confirm or rule it out as bad data. People shouldn't feel gunshy about this. (My field even has a conference for failed results, CFAIL.) I like to highlight examples of this stuff, because I see so many angry online comments when a paper turns out to be "wrong" or doesn't replicate.
Yep. Publishing like this gives a heads up to those operating similar observatories to keep an eye out for similar events. And it gives a nudge to theorists that might help them start looking in a more fruitful direction, appropriately caveated that it may be a statistical fluke.
You think that people's findings should be communicated by email? that their email chains are what should go into the permanent record and be cited and printed out and included in journals and such?
would you include all the quoted text in the reply-alls, or is that too much?
Yeah but no matter how hard I look I never seem to be able to read anyone else's email. Maybe OpenAI's upcoming models can help me find those references.
the fact that they are sometimes communicated by email doesn't mean that the best way to communicate them is by email. Personal correspondence references are specifically terrible as references since you can't go read them...
Kind of arrogant no? Linux kernel development mailing lists are producing something immensely valuable with a much clearer impact on economic indicators than your average scientific paper. Comparing them is hard, but it's patently absurd to say there's nothing being produced compared to scientific papers.
Dunno who you're arguing with, I didn't say they didn't produce anything of value. I said they aren't producing scientific papers. Conversations are not like papers. The scientists have conversations (sometimes on mailing lists!) as well. The analog to scientific papers in the Linux world are... scientific papers. And the occasional essay on the mailing list, which---get this---would be more valuable to humanity if it was subsequently reproduced as a paper with references and explanations and the like.
(Notwithstanding the absurdity of academic publishing, of course.)
Instead of reading everybody's spam or having to have a centralized body decide who gets to send messages to the mailing list, journal editors filter which things are worthy of publication and in which journal such that readers don't have to wade through garbage or uninteresting results.
If I had the email address of every researcher in my field, I would never send a mass email to them describing my latest goofy idea. I would, however, send my latest goofy idea to a conference with those same reviewers (if I felt it was technically appropriate and correct.)
Or this[2] 2007 Science paper on ultra high energy cosmic ray source candidates ("anisotropy") that we had to retract because significance started dropping almost the day the paper was approved.
It was a fascinating experience as a junior member to follow the collaboration internal conversation and investigation on this, because a lot of extremely principled scientists were clearly deeply worried about losing their hard earned reputation. In the end, I am convinced that we were simply unlucky.
Reminds me of the FTL neutrinos too, where the scientist where pretty much "hey, something is wrong, can you help us figure it out?" and the general public were the ones screaming "OMG! Physics is dead!"
Then when it comes out as measurement error, the public is all "Damn these scientists are all hype machine clowns..."
Unfortunately it wasn't just the public: it caused so much uproar within the experiment that two of the highest ranking members resigned their posts [1].
I was a bit dismayed at the reaction within the physics community. Experiments absolutely do need to follow procedures like blinding and careful internal review (especially before the data unblinding), but you can only spend so long designing the analysis before you unblind, and there are opportunity costs to cross checking everything. In an optimized community experiments will inevitably make mistakes. And once you unblind, it does no one any good to sit on an anomalous result forever.
I’m fine with that. Put it at the feet of pop science blogging.
I’m less fine with the time and resources spent on mouse models. They already know you’d get the same utility from a magic 8 ball, but they do it anyway.
Mouse models are useful to discard very bad ideas. There was a recent experiment to use bacteria to kill cancer https://news.ycombinator.com/item?id=46306894 They tried like 40 bacterias in vitro, then like 9 in mice, and only 1 was useful in mice and they will continue only with that, perhaps in humans. Anyway, as you suggest, there is a high chance it will fail.
Also, you can do nasty stuff to mice that would never be allowed with humans. In that experiment they injected cancer cells in mice with a bad inmune system, so they could get like 90 mice with cancer and run the experiment in a short time. No ethical committee would approve that in humans.
As another comment points out: that's nothing compared to high energy/particle physics!
The Pierre Auger Observatory certainly is a large collaboration for the astroparticle physics domain though. It's a big international collaboration.
Quick anecdote: my name (S Mueller) is not on that paper's author list because we had a rule that you had to be in the collaboration for a year before getting authorship. You stayed on for a year after leaving. Very reasonable! At the time I was nonetheless a bit bummed about missing out on the big Science paper. I guess I'm on the retraction though ;)
> That said, particle physics history is full of 3 sigma particle "discoveries" that disappeared with more data.
The idea is that because 3 sigma means a ~1/1000 chance of the thing being explained by random chance, 1 in 1000 experiments will produce a bogus 3 sigma result, and we do many thousands of experiments.
and also because a result always has the caveats of "if we did our experimental design and math right". A 1/1000 rate of experimental/code design errors will double the number of incorrect 3 sigma results.
In one sense anything that passed all their background rejection is a WIMP. To interact with a nucleus through so much matter, it's not interacting electromagnetically. The main candidate for a strong force interaction would be a neutron, and they did a lot of work to model that and eliminate it as a background. So definitionally it would be a WIMP.
They were pretty model agnostic in what they were looking for. They modeled and simulated a number of different ways a WIMP could interact with normal matter. If this is a discovery, more data will be needed to figure out the nature of that interaction and how it fits into particle physics.
But there's always a chance it's something completely new, or some extremely rare manifestation of things we already know about, but have never seen before. And even if it is WIMP, it may not be the right type of WIMP (wrong mass, or wrong interaction strength) to explain cosmological dark matter.
The detector from which data is taken to do this analysis contains 7 tons of liquid xenon. It is inside of a larger detector, which contains hundreds of tons of water and more than 10 tons of a scintillator. One of the functions of that outer detector is to absorb neutrons and other infiltrates coming from the rock.
When doing this kind of thing, the analysts will plot the rate of events as a function of "distance from the outer surface" and confirm that it decreases, and in this case "is 0" inside of the detector used for analysis.
But keep in mind that all statements are made statistically, so it's not that the event _can't_ be an external neutron, but that it is _very_ unlikely to be.
They have a lot of shielding for the detector, and also if there were a significant source of neutrons they'd expect to see other evidence, both in the detector and in the separate "veto" detector that surrounds the main detector.
That's not to say it can't be a neutron, but it would be surprising if it were.
The mainstream TV news report that I saw about this ended with a comment about how we should continue to fund this detector. Made me wonder if the nature of this release involved forces other than purely scientific ones. Apparently funding has already been cut for the successor to the LUX-ZEPLIN detector.
Particle physics is not a particularly large community. There is a hand-countable number of experiments like this, and the folks working on each of them know the folks working on each of the others. The collaborations executing each experiment are comprised of scientists employed by multiple independent institutions, both public and private, typically across national borders. Internally, the collaborations have a democratic structure with individual researchers acting as institutional representatives serving in what is essentially a parliamentary structure to make decisions. The software to determine results is always public within the collaboration and reviewed well in advance of making any truly public disclosure like this.
Culturally, an attempt to intentionally distort or misrepresent data to suggest a result like this would not be tolerated. You can imagine a bad actor writing a single-author paper with fakery, but a collaboration at this level is inherently critical of itself and everyone holds each other accountable.
But mistakes do happen. Exciting results or hints of exciting results can appear due to well-intentioned researchers making convenient mistakes which get glossed over for psychological reasons and then add up to appear as something significant. You can read about "blinding" (which is mentioned in the paper) to get a feel for the techniques researchers employ to not only guard against fabricating results intentionally, but even unintentionally.
The small pool also lends itself to a lack of unbiased reviewers. If everyone in the community is more-or-less tied to the research, who can best objectively test it, from a blinded POV?
That is a great point. I'm not sure if I fully understand your question, but I'll comment on both "testing," meaning doing the analysis, and "reviewing," meaning peer review.
The point of "blinding," as I mentioned it before, is to guard against biasing the result due to choices made by analysts while figuring out how to compute their final answer. Part of that is just psychological --- if an analyst _knows_ that the data is obscured, for lack of a better term, in a way such that they can't believe a signal that they see (for example --- that's just one way to blind), then they won't feel any emotion or excitement about things they encounter while constructing the analysis, because they know what they're looking at isn't real, and that helps them to maintain objectivity. It's also common to have multiple independent analysis teams for something like this --- they know that they'll have embarrassed themselves if they don't all get the same result both before and after unblinding, which provides pressure to maintain objectivity.
As far as review goes --- there are physicists who believe in dark matter, and there are physicists who don't believe in dark matter. But all particle physicists and nuclear physicists use the same core technical and statistical methods. For a result like this, there will be a few "core analysts" who do the analysis, typically graduate students and postdocs. Their advisors review their work, through the lens of making sure that they are doing so sanely. The collaboration will also establish an internal review committee to comb through the documentation and software which produced the result, not necessarily trying to disprove what they found, but really looking at it critically. Once it gets to a journal, they'll get reviewers who may be biased for or against the result, and who may or may not work directly on dark matter detection, but who can all assess the methodology by which the result was obtained. That's probably the best we can do with humans involved.
I'm not suggesting distortion or misrepresentation of the published result. The actual preprint is quite clear on the nature and limitations of the result, and there's no reason to think that what they're reporting didn't happen as described.
However, the context here is that the DOE has already paused funding for the experiment's successor, XLZD, for an unspecified amount of time which realistically, is probably going to depend on the next US presidential election. At the same time, the DOE encouraged the LZ project to develop scenarios for continuing to operating beyond its current end date of 2028.
Now, here's what the NBC Bay Area report I mentioned[0] ended with:
> "All the more reason, they say, to keep these machines running. [...] They'd like to keep the machines running, they wanna keep doing this research, they're applying for funding now, so fingers crossed, we'll have more updates on this [...]"
The media attention this received was not the organic result of some journalist noticing the preprint or the talk at TeVPA in Japan. There was a major wave of synchronized official press releases, coinciding with the TeVPA talk, from Berkeley National Lab[1], Brown University[2], SLAC[3], Brookhaven[4], Stanford[5], University of Sydney[6], and others including UMass Amherst and Imperial College London.
Now, these institutions are all involved in the collaboration somehow, so it makes sense that they would coordinate press releases for a major result. The question is whether this result warrants such treatment. It's a single event at 2.6 sigma global significance. Promoting it in this way was a choice, and I'm pointing out that it seems quite possible - in fact I'd say extremely likely - that that choice was made with the funding situation top of mind.
Thanks for clarifying, and my apologies if I came off as argumentative --- what you _are_ suggesting makes sense and is a reasonable thing to wonder about.
But I don't see anything particularly strange or coordinated is happening. From the collaboration's perspective, it's quite stressful having this event. They've already unblinded, so it would be unethical to do anything other than report what they found. If they publicize it and it's a mistake, then that's a big blow to their credibility. But if they withhold it and it's real, then they miss making the discovery and/or bias their future analyses on larger datasets without disclosing that to the community. So they are in a tough spot, and are safest to just tell the world what they saw.
This is getting media attention because it would be a big deal to the general public if this ends up being a real. Someone in the field wouldn't claim that it's real, but the possibility is catnip to folks looking for a sensational headline. That the press releases are synchronized in time is because the result was just released right now and they're all doing their commentaries right away --- for something like this, each institution independently negotiates a release with the local researchers who are involved. They all promise to wait until the result is officially released, out of respect for the scientific process, like you say, but the different institutions aren't coordinating with each other. They're just all respecting their own researchers.
Is a 2 or 3 sigma fluctuation worth a lot of press? Personally, I don't think so. But we don't know if it's a fluctuation yet, and no institution is going to pass on having made it clear, if this does turn out to be real, that they were involved.
All that being said: I would very much like these folks to continue to receive funding. They are professional and do excellent work, as demonstrated here.
> They've already unblinded, so it would be unethical to do anything other than report what they found.
Absolute and utter ridiculous nonsense. Can you not just admit when you're wrong?
It's perfectly ethical for them to give a talk at TeVPA about a paper they've published.
But at least 8 global, coordinated press releases? That's a choice, with consequences.
And one of those consequences is that they reveal themselves as chasing funding above all else. Scientific rigor goes out the window. 2.6 sigma results become amazing new discoveries.
It's not really their fault - it's systemic. But don't try to pretend that this is somehow the normal process of science being conducted with integrity.
It's mostly about signal-to-noise. The drawback to those detectors is that because the nuclei are heavy, when the neutrino interacts with them, they don't end up moving very fast and so there isn't much of a signal produced in the detector. They have to be cooled to cryogenic temperatures to even have a chance of seeing a coherent scattering event above thermal noise. Yes, the interaction cross section for the neutrinos is larger, but you can't pick out individual events very easily.
The 2017 paper you linked is the first observation of the process. They had to do it statistically; there's no smoking gun event. It took them almost a year with the thing sitting next to a neutron beam to get enough statistics. With a beam, they were able to do extra noise rejection based on the beam timing. It's still a lot of experimentation and engineering work to go from that to something that can operate in a lower signal-to-noise environment.
Also, growing large scintillator crystals is a very specialized process and so they are expensive. The cost is going to limit how large you can make such a detector. You can scale up a water-based detector much more easily, and that extra mass can make up for the reduced cross section, depending on what you are trying to observe.
The name is misleading. The glyphs are showing individual chord shapes. I can't write out a song using this. At best I can use this at the top of a tab to remind myself how the chords are meant to be shaped. But that doesn't appear to work much beyond the basic cowboy chords. For example, I tried 577655 which is an A major barre chord, and it didn't render. I realize a font can only do so much, but I wouldn't pay for this.
"VexTab is a language that allows you to easily create, edit, and share standard notation and guitar tablature. Unlike ASCII tab, which is designed for readability, VexTab is designed for writeability."
all textual representations of data are "formats" and one being easier to edit is a totally valid use case. like 'x57675' instead of a full tab. or # title instead of <h1>title</h1>
Not really. The lilypond format is extremely...complicated and obscure (and that's my polite take on it). Even simple stuff is quite complex. Very very far from Markdown's virtually WYSIWYG.
Yes, but music notation is not simple text. Markdown was made for simple text. If you try to represent more complex text, it turns into Asciidoc or Rst.
yeah this is not what guitar tabs are and I don't think a font should be used to do it or is the best method to do it. It can get really messy with time signature changes and managing all the strings and marks etc just by text and a font
I was excited for a second, because this is one piece of the puzzle (chords), then numerals solve melodies (you can just type something like 0-1 for open string first string, etc), then just need something for ornamentation. Seems like it only matches against a known set of chords, though.
I said to a few friends that the recent trailer felt like it could be for a House of Leaves movie. Different overall setting, but the "found footage" aspects, and the narrative over it, felt like they could be right out of the Navidson Record.
I don't have any real proof for this, but it feels like House of Leaves inspired a lot of the people making "found footage" and "creepypasta" stuff one the internet in the 2000s and early 2010s (SCP, Marble Hornets, Slender Man), and then that stuff came together to inspire the Backrooms.
The term I've heard for this sort of thing is "Physical Neural Networks" or "PNN"s. My impression is that one of the big things holding them back is that because we can't manufacture components to perfect tolerances, you can't train a single model and reuse it like you can with digital logic. Even if you can get close, every single circuit needs some amount of tuning. And we haven't worked out great ways to train them.
There's a lot of research going on in this space though, because yeah, nature can solve certain mathematical problems more efficiently than digital systems.
The "figure out what you want to say" is key. I've started to think of LLMs, at least in a business setting, as misunderstanding amplifiers.
How many times at work have you been talking to someone else where they're using common words as jargon? Maybe it's something like "the online system" or "the platform". And it's perfectly clear to them what they mean, but everyone else in the company either doesn't know what that actually is, or they have a distorted idea based on the conventional definitions of the words. Even without LLMs in the mix, this can lead to people coming out of meetings with completely different understandings of what's going on.
My experience is few people are actually providing the relevant context to the LLM to explain what they mean in situations like this. Or they don't have the actual knowledge and are using the LLM in the hopes it'll fill in for their ignorance. The LLMs are RLHFed to sound confident, so they won't convey that they don't know what a piece of jargon means. Instead they'll use a combination of the common meaning and the rest of the context to invent something. When this gets copy/pasted and sent around, it causes everyone who isn't familiar to get the wrong idea. Hence "misunderstanding amplifier".
To the point of the article, this is soluble if people take the time to actually figure out what they are trying to convey. But if they did that, they wouldn't need the LLM in the first place.
And that people and the systems actually know the relevant terms.
I recently was dealing with the Amazon robot--after correctly identifying the items in the order it then proceeded to use short terms which were wrong, but make sense as what a classifier might have spit out. Instead of understanding being a shared thing it falls entirely on the user. Sufficiently adept user, this is fine. But a lot of users aren't sufficient adept.
This doesn't seem to be complete. It's missing the Waste Isolation Pilot Plant, for example, which should be southeast of Carlsbad, NM. It's a underground salt (metal/non-metal) mine, and MSHA definitely regulates it
reply