I'm perfectly OK with papers that explore new directions without apparent practical applications. That's not what I'm asking about. I'm asking about papers that present a new way to encrypt digitized image, when that's a solved problem; give security arguments that do not hold water; and, for many, that have claims so vague as to be un-falsifiable. They go by the thousands.

@fgrieu I'm not sure I see the distinction between "new directions without apparent practical applications" and "new way to encrypt digitized image". I certainly agree that there are many papers without merit, but the general notion of considering a new approach to a "solved problem" is not meritless. Are there researchers at top institutions writing papers on this subject? If so, do you feel the same way about their papers?

Would you be able to provide the pair of papers you reference? Those both sound really fun.

@fgrieu So who's going to judge whether research has any practical applications or might lead to some meaningful future results? Ultimately that's going to be very subjective and prone to error. Also, you have to draw the line somewhere, sure, but trying to limit research strictly to things that have some clear potential benefit is probably going to do more to harm innovation and invention than anything else (but this is not the same as saying you should reward research that has no apparent purpose).

@StevenGubkin: I don't know which papers Andreas Blass refers to, but such a proof of the Lagrange inversion theorem can be found in "Functional composition patterns and power series reversion" by George N. Raney (1960). pdfs.semanticscholar.org/3792/… You can find discussion of this in Flajolet & Sedgewick in item "44 Conjugacy principle and cycle lemma" and the rest of its page (where they refer to "Polish notation" as "Łukasiewicz codes") at algo.inria.fr/flajolet/Publications/book061023.pdf#page=81

@StevenGubkin Unfortunately, I don't have references for these papers. I came across them back in the 1980's, when I was working on the set theory volume of the Omega Bibliography of Mathematical Logic (classifying papers according to subject). The papers and bibliographic data are surely in that bibliography (though not the set theory volume), but without abstracts, they'd be very hard to find.

-1 for not answering the question, except in an indirect and barely relevant manner.

@AndreasBlass A better example in mathematics would be the flood of trash papers claiming to 'disprove Fermat', 'prove Collatz', or much worse, 'disprove mathematics'. A key difference is that these are usually not funded by (mathematical) institutes.

@Kimball : "Are there researchers at top institutions writing papers on this subject?" : no. The accepted wisdom in academic crypto, since the 1990's at least, is that, schemes (encryption, signature...) operate on arbitrary bitstrings. Specializing that is plain not worth interest. When it come to application of encryption to any specialized field, things truly boil down to: compress unless there's a domain-specific reason that compression ratio leaks useful info (as can be the case with voice), then encrypt with a general purpose algorithm. Nothing else is academically recognized.

@fgrieu do you have a proof that the image encryption is a totally closed subject with no new insight possible? You don't. And The Wise you refer to don't have it as well. It is just an opinion. And arguing what is more and what is less important is pure philosophy, and IMO the waste of time. Why does it bother you who publishes and what anyway?

@freakish: I'm bothered that these articles are written by students working on crypto that missed some of the standard conclusions of the field: (a) security of an encryption algorithm can and should be independent of the kind of data encrypted (b) security analysis can't be made with automated tools. I'm not against working on esoteric subjects and unconventional tracks, but only after getting the basics.

@fgrieu both your statements are purely opinion based. Plus I strongly advice to care less about what others do, especially when it doesn't affect you (or actually anyone) at all.

@fgrieu how does one encrypt a VHS tape onto a VHS tape? There's a subfield of encryption whose name I don't remember, which is about encrypting phone numbers so they still look like phone numbers, addresses so they still look like addresses, etc. Even though encryption in general is solved, when you add additional requirements that make the general solution inapplicable, it is not.

@freakish : I welcome you to ask about (a) and/or (b) on crypto-SE so that I, and others, can give you arguments, including near-mathematical proof for (b). Bad articles, which nowadays tend to quantitatively dominate the field of crypto, are a nuisance because some people trust them (perhaps: are told to work on them) and do not study actual science. I'm proud I care about preventing that. Also it creates unwanted noise on crypto-SE when people ask questions about these articles, and we have to debunk them.

@user253751: you are right about VHS, which is not digital data: it requires special encryption means. You right also that adding extra constraints to encryption of digital data (like the ciphertext of a phone number looks like a phone number) requires special means (Format Preserving Encryption). Yet the was majority of articles on image encryption adds no constraint that standard encryption does not have.

@benrg no, I don't believe there is such thing as objectively poor work. The fact that you say "believe" already convinces me there is no such thing. And again: some people like painting. I personally find this activity pointless. But why would I spent time debating this? Or worse: preventing them to paint? Very dangerous mentality.

@fgrieu I really don't care about "near-mathematical proofs", which is just another way of saying "we strongly believe". Once you get the real proof let me know. Otherwise all of that is just whining to me.

@freakish: That's correct: the proof of (b) does require (and easily follows from) hypothesizing the existence of computationally secure encryption, per the academic definition of that (e.g. in Jonathan Katz and Yehuda Lindell's Introduction to Modern Cryptography, 2nd edition, definition 3.7). Which is not certain, in particular because it would be impossible if P=NP, which is not demonstrably ruled out.Thus if modern cryptography is not nonsense, security arguments using statistical tools operating on the results of encryption are nonsense.