whuber

@StephanKolassa You either need one or you don't--there's no other option ;-).

@StephanKolassa Done, thanks.

@Scortchi That's what I had concluded, too.

@Scortchi Right. I'm torn between profile editing and account destruction. Not wishing to make a system-wide error, I have conservatively settled for destroying these accounts here on CV. I'm averaging about four per day, ranging from zero to seven. I'm sorely tempted, though to edit the profiles more meaningfully, such as by asking readers not to patronize these businesses!

@StephanKolassa It happens about once a year on average, roughly evenly split between legitimate research and crank publications.

@StephanKolassa Alas, after a pause on Monday, this activity (evidently based near the Indian subcontinent) resumed yesterday: I destroyed seven accounts.

@User1865345 re Poisson: See Stigler's Law of Eponymy.

@User1865345 Thanks for letting me know. Because I have suspected some intermittent failures of the formatting mechanisms I haven't bothered to track it down when the formatting doesn't show.

One possible (speculative) conclusion is that the chatbots have acquired most of our material and are feeding it to people who otherwise would come directly here with their questions. In other words, the interest might be out there, but--like vampires--"AI" is sucking users away. That's a terrible thing for SE but maybe not so bad for CV's most active users.

The sole exception is a huge spike in new users (having no meaningful interactions with the site) occurring with the advent of ChatGPT et al about a year and a half ago.

@Glen_b On many measures, we're operating at 30% to 40% of our peak in the late teens.

@J-J-J Okay, it's open again: go for it!

@User1865345 There used to be many questions of this sort on the Meta site in the early days. Based on my recollections of the answers, it's likely your score was reduced because a user who voted for one of your answers was removed. The system does not retroactively un-award badges when that happens.

@StephanKolassa Done, thank you.

(It's BS because energy changes are commonplace--they accompany virtually every interaction in the universe--and do not require or even suggest any application of negative probabilities, even when (as is usual) a Fourier Transform of a spectrum is analyzed.)

@User1865345 It's not wrong--but it's not relevant (arguably, it's a lot of BS) and it's certainly not a "layman explanation"!

@User1865345 Re "takes ages:" It depends on which mods are active at the time. A couple weeks ago I happened to see one of these accounts pop up and destroyed it before they even had a chance to post anything. That seemed to stop them in their tracks--for the day, anyway... .

@J-J-J Of course you can flag users--but first consider using a comment to provide friendly advice. A user who is familiar with the site (or SE generally) yet persists in repeated destructive or obnoxious behavior ought to be flagged for moderator consideration. Treat new users gently, please.

@gung-ReinstateMonica Yes, indeed. And often it's a strong hint of a poorly formed question with no research. I have been using that tag for years to make easy, quick closure decisions ;-).

@User1865345 One individual (or group) has been spamming many SE sites in this fashion for several months. The harder they try, the better the system gets at identifying them. We don't have to worry about such attacks. We are better at unearthing the subtler (and more patient) trolls: that's where your flagging efforts are the most helpful.

@J-J-J I confess I am unsympathetic to requests for generic "real-life datasets." There are millions of them out there, easy to find, freely available, in any discipline one can think of.

@User1865345 Even more remarkably, over those ten years the OP visited this site only 33 days: not a frequent user by any means--but, evidently, somebody who checks in and knows how to participate.

@psie "Infinite-dimensional vector spaces" are more properly the subject of analysis. A good textbook on linear algebra will clearly indicate which theorems apply with no or simple modifications to all vector spaces and which ones are limited to finite-dimensional spaces.

@J-J-J Unfortunately not. Indeed, it seems to be the exception rather than the rule that the posted close message accurately conveys the close reason. What you did by commenting is about the best we can accomplish.

@Galen Thanks for the reference. It sounds remarkably like Mario Bunge's approach in his Foundations of Physics (1967). I read that at an early age and am sure it influenced my thinking about probability.

@SextusEmpiricus A standard textbook on analysis (Rudin) systematically develops a theory of complex valued measures. Negative measures are no problem!

In that sense, you can find examples here on CV, such as stats.stackexchange.com/a/72486/919 (sums of Gamma variables expressed as mixtures of "extraordinary" variables).

"'this cannot really be interpreted as the simultaneous probability for coordinates and momentum' but can be used in calculations as an auxiliary mixture measure."

@NickCox Polson and Sokolov describe elements of that paper. Interestingly, they quote Heisenberg concerning negative probability (densities):

@SextusEmpiricus I try to make comments that move the conversation forward. And I also try to conform to the rules of this site, which ask that comments not be used to provide answers.

@User1865345 I have been known to do the same thing (misunderstanding)... . I think many of us have done that.

@NickCox I have been unable to contribute to other stats fora, such as those on LinkedIn and even on ASA community sites, because they all suffer from the same problems. The lack of moderation and voting dooms them even when all the readers are professionals.

Reference: Goldman, W. 1973. The Princess Bride. See: "Miracle Max."

The epitome of wry understatement: "For a survival curve to increase, people would be coming back to life some extended time after being recorded as officially dead. This is regarded as unusual." Glen_b, 2025-01-19

@NickCox (I fixed the unfamiliar/unfair issue for you.)

@J-J-J Math and MO are (very) different sites!

@User1865345 That had already been posted eight hours earlier (by another user) at stats.stackexchange.com/questions/660191.

@NickCox The Athenians killed Socrates for failing to honor the gods and for corrupting the city's youth, but that was pretext. In the Platonic dialogs he invariably and quite publicly demonstrated that prominent citizens (and peripatetic sophists) claiming to "know" a concept (such as goodness, wisdom, knowledge, and so in) did not know it at all.

@Scortchi You (and others) will be interested in the "Users destroyed" stats in The Year in Moderation 2024. They're consistent with your 6-12 profiles per day (2200 - 4400 annually).

@GuillaumeDehaene It looks like there might be several misconceptions lurking in your question on Math, such as assuming $C$ even exists and supposing that $C^\infty$ functions are necessarily analytic. I suggest consulting a complex analysis text for background. That likely will indicate the answers but even if it doesn't, you will be able to reframe your question without involving dubious assumptions.

Alternatively, they could have spent a couple of hours reading related threads here on CV and discovered many of the basic problems with that question, as well as obtaining many answers.

@NickCox The original version of that question was asked so many years ago that the OP could have spent just a small time learning some prob & stats in the interim and become an expert in the subject by now ;-).

The ones that are important to flag are those that interact with the site in some significant way, such as posting stuff, voting strangely, etc.

Interestingly, all four are from blacklisted IP addresses. That implies SE already knows these are problematic and we likely don't have to do anything (in these particular cases).

@StephanKolassa It is helpful. Consider flagging them, which you might find to be a more convenient mechanism.

You've just taken yourself right back where you started. Evidently I cannot be of any help to you--sorry.

Now you're getting somewhere. But that's a lot of questions. You need to think about which question to focus on first. Take your first question about prediction: what data do you have about this (hypothetical?) absentee that could be used to predict a score? That will have to be related to the data you have about the test takers themselves. Describe those data: what they represent, how they were measured, and (possibly) their statistical characteristics.

The key is to explain, in plain and simple terms, what your problem is. It is not about "modeling likelihood" -- that sort of problem is only for academic researchers who understand how to do such things. When you're totally lost, don't present any math, don't speculate about what the answer might be. It looks like you want to analyze some test scores. For what purpose? What do you want to learn about them or what decision do you want to make? Why don't any of the many posts about test scores here on CV suit your needs?

The first paragraph is such a mish-mash of separate, nearly unrelated ideas (Bayesian analysis; "filtering;" robustness; modeling; maximum entropy; "principle of indifference;" and bias) that it suggests (a) this post needs much better focus and refinement and (b) it might be an XY problem where your proposed procedures might have little to do with what the underlying situation it is that you face.

This looks like a version of your previous question.