Wednesday, August 22, 2007

Withdrawal

While blogger was down most of the morning, I recalled a recent conversation about the arXiv's policies. Since not all of the readers around here are registered arXiv users it occurred to me that you might not know it's not possible to withdraw a paper from the arXiv. However, I have good evidence that most scientists are indeed human and occasionally make mistakes - and not in all cases realizing a problem is identical to solving it. The only way to deal with that on the arXiv is to replace the paper with an empty update, or with a dummy file. The previous version(s) of the submission however, remain available on the arXiv.

This happens more often than one might think. I just did an abstract search for 'withdrawn' which produced the warning 'Your query resulted in too many hits, only 1000 hits are being displayed.'.

I don't particularly like this. Nobody is eager to admit mistakes, and replacing a paper with the comment 'withdrawn' means it will appear in the next listing under 'replacements' with a clearly visible stigma. I think that probably some people who have realized a problem with an argumentation prefer to just leave the paper as it is, and hope nobody ever mentions it again. The result however is that weak or wrong ideas remain on the arxiv, publicly available. Yes, it's called a pre-print server, but nobody is required to add a comment saying: This paper has been rejected by 6 different journals because I've abused infinities (see previous post) and I've eventually understood why.

That is to say, for this reason I would think the arXiv quality might improve if it was possible to withdraw a paper - completely - from the listing.

However, on the other hand I am afraid it might encourage users to submit premature works, knowing that they 'could' withdraw if it turns out to be bullshit. Maybe public humiliation is the better tool to guarantee quality? And I guess I would miss some of the stage fright that comes with pushing the 'submit' button (and don't yet relax).

So, I am kind of undecided on that issue, what do you think?

39 comments:

stefan said...

Dear Bee,


incidentally, when driving to work this morning I heard on the radio a report about a changing "error culture" in companies, and that in most cases, it may be beneficial for everyone if people are allowed to make errors and to learn from them, instead of trying to avoid them at all prize - and to sweep them under the rug or to blame them on other people, in case they happen.

So, in general, replacement of papers with a statement that this and that error makes obsolete the previous version of the paper seems to me to be not a bad idea - at least, you can keep track of what was going on.

Complete deletion of papers with errors may be an option, but obviously, this may lead to broken references to some paper arxiv:XXX which has been deleted. I guess that may be the main reason that complete deletion is not intended.

This is in fact a more technical issue, but there is a similar problem with "official" papers on the online archives of journals: You could imagine that you may replace a PDF file if you have spotted some typo or so - not to speak of any errors in actual content. But that does not work once the files have obtained a DOI - according to the regulations, some file identified with a DOI has to remain unchanged forever. This rule seems to come from the time when what was printed on paper was printed and could not be changed anymore - but on the other hand, it makes sure that if you cite some paper, you can be sure that whoever will sometime in the future look up the reference you have given will end up with the same paper that you had seen when you quoted it. It just avoids confusion.

It may be that a complete deletion of papers would improve the quality of the arxiv - but on the other hand, the arxiv is some kind of record for all kind of ideas, also those that are wrong. And you do not remove papers which turn out to be wrong from journals either...

Best, stefan

Kea said...

The fact that things cannot be deleted from the arxiv is its BEST feature, by far!

Uncle Al said...

http://arXiv.org/abs/physics/0205089

Table at paper's end. Turn a dial. A 60.5 kg lump exhibits 41 kg of buoyancy. Oh NASA! Anybody who posts such deserves its being immortal, e.g.,

http://arxiv.org/abs/physics/0108005
NASA + $(US)1 million = nothing

Robert said...

I think it's absolutely essential that you can see the full version history of a paper on the archive including content that was so wrong that the author could not correct it but had to withdraw it. Otherwise, you would not really be accountable anymore for what you place there. You could even think of submitting placeholder papers that are later replaced once you worked out the real content just to scoop your competitors.

Which reminds me of this strategy to get many citations: You write a so-and-so paper in a fashionable field but include some more or less obvious technical error. Then all nit-pickers will refer to your paper just to point out that error.

Bee said...

Hi Stefan,

I was not talking about published papers. If a paper has went through a peer review process and was published with DOI that's a completely different case. I totally understand the argument that the availability of a mistake made can be useful. However, I happen to know of several cases where I have had discussions with authors, and where the authors after an exchange admitted that assumption soandso is unjustified and the result doesn't make sense. All of these papers remained on the arxiv in exactly that version. Even though the author knew, and admitted, that the conclusion doesn't make sense. I just don't think this is good, and I am wondering whether a reason for that might be that the present way to 'withdraw' a paper creates attention. Besides this, not every mistake is a mistake that needs to be documented because one can learn from it. Best,

B.

Bee said...

Hi Robert,

yes, that people wouldn't feel accountable for their submission as much as they do now is also what I would be afraid of. However, when you say

You could even think of submitting placeholder papers that are later replaced once you worked out the real content just to scoop your competitors.'

this is not necessarily the same thing. One could keep the whole history with all new versions, unless there is a withdrawal, in which case everything is deleted (including the number which gives you the 'I-was-first' right).

Is having something on the arxiv (i.e. not peer reviewed) sufficient to claim 'I-said-it-first'?

Best,

B.

Best,

B.

Bee said...

Hi Stefan,

It may be that a complete deletion of papers would improve the quality of the arxiv - but on the other hand, the arxiv is some kind of record for all kind of ideas, also those that are wrong. And you do not remove papers which turn out to be wrong from journals either...

No, but if it was published you publish an erratum, and if it was rejected nobody ever hears of it. The arxiv is not a print journal, but it's also not a pre-print journal anymore. It's turning into a never-to-be-printed journal. Those of us who check it every day, or who use abstract search etc, have to sort through an increasing soup of stuff that nobody would miss would the author be willing and able to withdraw it. Best,

B.

Anonymous said...

Incidentally its rather rare for papers to get debunked on arxiv. It happens for high profile papers, or ones where the error is not very apparent, but by and large the immense majority are left intact or simply ignored.

That was sort of the hope with blogs: Namely, that highly speculative papers or ones that are out of date could be linked to a comment section where people discuss it technically in such a way that it doesn't require a full fledged paper to spell out ad nauseum the inherent problems.

-Haelfix

Bee said...

Hi Haelfix,

Yes, one could have hoped that. But I find it rather unlikely to work. I have suggested previously (see e.g. here) that the arxiv allows comments to papers that don't have to appear in the listing as a new paper, but can be used as a discussion to an existing paper. I.e. if you would look at a paper, you could immediately see comments to it, who left the comment, plus it would be arxiv users only which I'd think would remove most of the background noise one has on blogs. Best,

B.

Eric Gisse said...

On the other hand, what does it tell you when you see a paper has gone through 20 revisions [literally] and is still crap?

I mean, other than "laugh your ass off". After that.

Anonymous said...

feqehcv

Arun said...

Dear Bee,

One more of those problems where there is a landscape of solutions.

Best,
-Arun

Neil' said...

BTW, how can a non-affiliated writer get something up on ArXiv, assuming it deserves to be there of course? Just find someone out there to sponsor, or can someone at ArXiv consider it? Are the people (mostly grad students?) running it really as snooty and inaccessible as critics say? I wonder if there are institutions, gatherings of those who can pool resources, look at each other's papers, that some sponsors are working with. tx

Prof Anonym said...

Bee asks: "Is having something on the arxiv (i.e. not peer reviewed) sufficient to claim 'I-said-it-first'? "

Absolutely. Whether we like it or not, and whether people admit it or not, the arxiv *is* the literature nowadays. Nobody reads journals, and nobody really cares whether something gets published eventually [except administrators].
If one needs to judge the quality of a paper [other than by actually reading it --- but who wants to do that? :-)] one does it by looking at the name of the author and his/her affiliation, and, in the case of older papers, citation counts. NOTE that I am not saying that this is good.

The truth is that what people look for are papers that might help them to write more papers themselves. It's exactly like evolution: the "organisms" that survive are the ones that are most successful at "reproduction". Write a dull paper in a popular field, but leave a lot of hooks dangling for people to attach themselves to, and you will be cited. The problem is that this leads to mediocrity: T Rex goes extinct, and is replaced by the common cow.

Actually I am not too sure what is bothering Bee. Sure there are a lot of bad papers on the arxiv, but there are and always have been a lot of bad papers in the paper journals too. The bad papers will be forgotten. Actually, a lot of good papers also get forgotten; this is much more serious. But allowing bad papers to be exterminated won't help to solve that problem, will it?

All I can suggest is that people, especially people with blogs, spend more time drawing attention to really *good* papers, and otherwise doing whatever they can to try to ensure that these papers don't get flushed away by the tide of the arxiv. I see an awful lot of blogs where bad papers get attacked, much fewer where good papers get celebrated.

arivero said...

Journals had asked for removal of the preprints after publication of the article, if it had been a initial option in the arxiv. Book editors could suggest authors that removal of previous lecture courses will improve the distribution of the book. Only the mediocre papers would remain.

CarlBrannen said...

neil,

I'm an unaffiliated author and got something put up on arXiv. You have to get a sponsor, which is not difficult if you know someone who is working on something similar.

After getting a sponsor, you upload your paper, and you are given a temporary number. If the moderators allow your paper, then it goes on to arXiv.

I had trouble with the moderators, but a few months later, I started picking up citations, and they agreed to allow it. Being fickle, I had a change of heart and decided to keep it on my own personal website where I can track downloads and update (or withdraw) it whenever I like.

Lumo said...

I think it's correct that one can access the previous versions because otherwise history would be constantly overwritten which would cause a lot of confusion and bad mood.

I think it's obvious that if a serious mistake is realized by the authors and they have this technical opportunity to fix or remove the paper, they should do it.

Unfortunately, this elementary honesty combined with self-reflection is missing for people who do things like DSR or LQG. The easy way to see it is that if these people were honest, all papers on these subjects would have already been withdrawn.

Kris Krogh said...

I agree with Kea that keeping all versions of an ArXiv paper is a REALLY good thing. I learned first-hand what sorts of trouble could happen without that feature.

A paper was posted earlier this year claiming a measurement of the frame-dragging effect predicted by general relativity, to 0.5% accuracy, using orbital data from the Mars Global Surveyor spacecraft. (Better than the expected accuracy from the Gravity Probe B experiment, to be announced later this year.) The author has been considered one of the top experts on such measurements.

I had staked my future on a different prediction for frame dragging. So I spent a couple of weeks looking for a fatal mistake in my own work. But when I finally checked this "high-accuracy measurement" carefully, I started laughing.

First, the author comically misinterpreted the data he was using. (As confirmed by the authors of the NASA paper from which he took his data.) I also found he'd altered a key time period to make his result come out close to what general relativity predicts.

So I posted a rebuttal on ArXiv. (Mine was followed by two others.) I fully expected this person would retract his paper. Instead, he did this:

First he posted a new version of his paper (v6), fixing several obvious mistakes I'd pointed out, without mentioning my criticisms. Then he posted a new paper directed to me, written as though I had falsely criticized his corrected paper!

That wasn't smart, since ArXiv keeps all versions. I only had to revise my paper, pointing out that I was referring to his version 5. And by the way, look at these unattributed changes which appear in his version 6...

Needless to say, I'm grateful ArXiv doesn't erase replaced papers! I think it's really helped those in the field understand this situation.

(This person has been publishing more than one paper per month in peer-reviewed journals. Not quite as many as this guy.)

Bee said...

Good morning all,

It's kind of interesting that nobody of you seems to acknowledge possible benefits of allowing withdrawal. Only Lubos seems to be aware that the ethical and moral behaviour among scientists isn't always all that great.

Since Prof. Anonymous asked, what is bothering me is that these not-even-wrong and never-corrected papers remain not only the arxiv, but get cited. There can easily grow whole ghost-trees of citations building around wrong papers. Seriously, if you want to add a reference that shows whatever result you have also appears within some other approach, you will find some support for it SOMEWHERE. And who really reads the details?

This is bothering me, because the more useless information is out there, the more complicated it is to find out the useful one. Yes, of course you can say good scientists should be able to figure that out, and I think most of them are. But the point is, not everybody always has the time or the patience to read every paper. And realistically there is a non-zero possibility that wrong or shaky ideas serve as condensation for even more crap.

Please let me mention again that replacing previous versions without keeping track of it is a completely different issue. And I think we all agree that it is definitely necessary to keep the previous versions (reg. what was commented above, I am not even sure if everybody who uses the arxiv is aware that the previous versions are still available).

Best,

B.

Dylan Thurston said...

Bee wrote:

However, I happen to know of several cases where I have had discussions with authors, and where the authors after an exchange admitted that assumption soandso is unjustified and the result doesn't make sense. All of these papers remained on the arxiv in exactly that version.

I am shocked by this behaviour if it's really as described, and I think that the perpetrators should be publically mocked. (There's some embarassment in having a withdrawn paper on the arXiv, so your task is to make the embarassment of not correcting the error worse...) I don't think the arXiv itself should have much to do with it, though; perserving the history is vital.

I see some analogy to public disclosure of security vulnerabilities; without public disclosure, software producers have no incentive to clean up.

Eric Gisse said...

lumo:

You *are* aware you sound like a crank right now, don't you?

kris:

I can't remember the last time I opened a link on arxiv, read the title, and simply exclaimed "no!". The abstract made me laugh openly.

The paper is Not Even Wrong. The LAGEOS spacecraft were specifically designed passive craft for measuring the Earth's gravitational field. They and the laser ground stations couldn't push past 10-15% error in measuring the Lense-thirring effect and this joker claims 0.5% on *MARS* with the global surveyor? Give me a break. I'd crap myself if the MGS' orbital parameters were half as well known as the LAGEOS craft's.

I hope every journal he submitted that one to laughed him off.

Oh, and about your paper...

It is wrong.

Why? Your theory predicts dipole gravitational radiation.

If gravitational radiation existed for changing dipole moments, the solar system would have collapsed in upon itself several billion years ago. Plus the decay of PSR B1913+16 wouldn't obey GR to within 0.2% [arXiv:astro-ph/0407149v1].

Oh, and having you claim you can "derive" c by assuming that gravitational potentials superimpose despite repeatedly referencing objects which have large amounts of gravitational self-energy is rather amusing. There is also the funny part where you assume SR holds despite SR assuming homogeneity of space and time, which is totally thrown out the window with a gravitational potential.

What else? Twenty three revisions of your paper and you /still/ mis-spell Nordtvedt's name. Speaking of Nordtvedt, the Nordtvedt effect, it doesn't really exist. http://relativity.livingreviews.org/open?pubNo=lrr-2006-3&page=articlesu9.html

I could go on, but I have written enough.

Your paper is one of the type I was mentioning earlier. You have gone through twenty three revisions since 1999 and it is still deeply flawed. Nor has it been published anywhere, ever.

Eric Gisse said...

Cute.

http://relativity.livingreviews.org/
open?pubNo=lrr-2006-3&page=articlesu9.html

QUASAR9 said...

Bee, I think Stefan mentions the reasons and most obvious arguments against deletion.

Alas, it seems if you want no information loss, it also means no wrong and erroneous information is lost either. hmmm of course these can often 'wild goose chases' - but lets hope the universe is not filled with any 'red herrings' - or the world of particle physics with too many virtual particles, never mind infinities

Uncle Al said...

The Nordtvedt effect (lunar laser ranging) is a potent Equivalence Principle test. Return to Luna to plant a perfect field of optimized retroreflector arrays.

Can biology source a parity Nordtvedt effect? Earth masses 5.97x10^24 kg. Living matter masses 1.15x10^16 kg, mostly achiral water. Protein L-amino acids but D-sugars. A 5x10^(-10) active mass fraction - as good as Weak Interaction studies - nulls as meat and wood cancel.

Bee said...

Hi Quasar,

I know that Stefan and other commenters above mention the most obvious argument against withdrawal. I am just puzzled that everybody honks the same horn there? You say, it also means no wrong and erroneous information is lost either - the problem is exactly to trace which articles turn out to be wrong and erroneous.

If everybody who realizes a mistake in his paper would add a note on that, that would be fine with me, no problem there. This would indeed be the ideal scenario I'd envision. Then you'd have a documentation of dead ends, and you could follow up a misleading thought to see where and why it went wrong or didn't work out etc. Great.

But the point is that I think many people just don't do that, because it doesn't look good. Consider you have a paper on the arxiv that has gotten some citations (half of which are your own) etc. Then you realize there is a mistake in it, and it's one that you can't fix (at least not in a finite amount of time). If it was possible to withdraw with as little humiliation as possible I'd think it would encourage people to actually do so. That's all I'm saying.

Best,

B.

Bee said...

Hi Dylan:

I am shocked by this behaviour if it's really as described, and I think that the perpetrators should be publically mocked. (There's some embarassment in having a withdrawn paper on the arXiv, so your task is to make the embarassment of not correcting the error worse...) I don't think the arXiv itself should have much to do with it, though; perserving the history is vital.

Again, let me point out I agree that keeping track of a paper's history is necessary. However, I don't want the arxiv to end up being a many worlds version of peer-reviewed journals.

I can assure you that what I've said above is correct, and I have heard the most ridiculous reasons for such behaviour. One guy said it's out there anyhow and basically shrugged shoulders (so I imagine, since we communicated by email). Another guy said essentially: but there is much more serious crap out there that even gets published. In several cases people just stopped replying to my emails when they ran out of answers. At least two insisted that a bug was a feature (that included equations with wrong dimensionalities, or mistakenly believing the Planck mass is macroscopically 'large'), and in one case I ran into a gummy wall of repeated thanks for my valuable criticism (that however didn't have any consequences).

I myself don't want to waste my time on writing a comment on such papers (especially not after I have wasted time arguing about it, enough is enough), and I don't like to publicly 'mock' people. If someone is a good scientist and comes to understand he has made a mistake I expect (s)he reacts accordingly. This however, sadly does not always seem to be the case, and I am afraid this unfortunate behavior is supported not only by the present research atmosphere, but also by the way the arxiv deals with submission.

Imagine you have to write an application. Your new employer does an arxiv search and finds you have withdrawn a paper, maybe even two. Wouldn't that be the first thing he'd look at? Yes, I would think the experience of making and admitting a mistake can be a definite pro for a candidate. But I guess not everybody who is in a hiring committee thinks so, neither do people like to have a close up examination of their past mistakes. So what is the result then? You leave the paper as it is and hope nobody looks too closely at it. Which is, if you have some other, better publications likely to be the case.

So, I am wondering if this behaviour could be avoided if it was possible to completely withdraw a paper from the listing.

Best,

B.

Bee said...

Hi Neil, Hi Carl,

Regarding uploading papers on the arxiv. Since a couple of years the arxiv asks for endorsement of new users. I don't know how this process looks from the user's side since I've been registered at the arXiv before they introduced this procedure. By now I have occasionally been asked to endorse somebody's submission. I don't know how it works for students or so, I believe they will get endorsed by their supervisor or something. I have been told however, that this endorsement is only required if there are some doubts about the user, but that initially everybody can submit?


Hi Prof. Anonymous,

Absolutely. Whether we like it or not, and whether people admit it or not, the arxiv *is* the literature nowadays. Nobody reads journals, and nobody really cares whether something gets published eventually [except administrators]. If one needs to judge the quality of a paper [other than by actually reading it --- but who wants to do that? :-)] one does it by looking at the name of the author and his/her affiliation, and, in the case of older papers, citation counts. NOTE that I am not saying that this is good.

There is certainly some truth in this, but I don't think it is generally the case. E.g. I read a lot of papers that I can't really make sense of immediately, so I push the idea back and forth for a while and put it aside (unless it's really tremendously interesting for my present work of course). If I later see the paper got published in a peer reviewed journal this might give me a reason to consider it more seriously. It's just a way for me to figure out whether a paper is worth time spending on - and this is a selection criterion that (ironically) gets more important the more stuff there is on the arxiv. Also, if possible I try to read the published version and not the one on the arxiv because many people don't bother to replace with the published version, which occasionally has considerable improvements (there are indeed benefits to peer review).

I believe what drives most people to jump on arXiv papers is the present need to be up-to-date and to be part of the next fashion wave. For this you can't wait for peer review to be done. The problem that papers get judged on by author and affiliation is imho a serious one. But again I want to emphasise that this gets more severe the MORE stuff is out there because its INEVITABLE that readers filter the new publications on the arxiv SOMEHOW.

So you're sitting there in front of that listing, and you need some fast, easy criteria to get through all of that in a minimum amount of time. Some people might check for familiar names. I personally check for keywords (these however can change from one week to the other). This sounds harmless enough, but now you have to consider that this is eventually what sets the part of the field that people are familiar with, which they read, think about, and later cite. These criteria that people necessarily apply are of course influenced by the rapid increase of the number of people working in the field, and the all the stuff they write up. Here, as in other cases, the question how to structure and organize available information is crucial to its usefulness. And I think this a problem that isn't (yet) taken seriously enough.

Best,

B.

QUASAR9 said...

"If everybody who realizes a mistake in his paper would add a note on that, that would be fine with me, no problem there. This would indeed be the ideal scenario I'd envision."

Bee, I agree

The problem as Stefan pointed out is that then you may not wish to refer or link to a paper, because that paper could be altered ...

you are suggesting that if the alteration is a correction, then it should be beneficial to the positive flow of information.

And help introduce ammendments and get rid of errors on the original paper. That's fine with me.

"Then you realize there is a mistake in it, and it's one that you can't fix (at least not in a finite amount of time). If it was possible to withdraw with as little humiliation as possible I'd think it would encourage people to actually do so. That's all I'm saying."

That's fine, except then anyone who has a link to that paper would have to tidy their paper. Not such a bad idea
And it could leave certain people who thrive on criticising other theories 'high & dry' too - lol!

Bee said...

Hi Quasar,

The problem as Stefan pointed out is that then you may not wish to refer or link to a paper, because that paper could be altered ...

And as I've answered above, this is a misunderstanding, I'm not talking about changing anything about keeping and tracing the paper versions on the arxiv unless withdrawn. Papers submitted on the arxiv can be altered already. What you get if you just enter the number of the paper is always the most recent version, unless you explicitly refer to an earlier one (but usually people don't specify the version number of a paper.) Occasionally authors modify their paper - yes, even the conclusions - but this is already the case. It doesn't seem to me that people are bothered by this. I wasn't thinking about changing anything of that. If somebody is afraid a paper might be considerably altered or withdrawn, I'd think he shouldn't refer to it to begin with (unless it's an explicit comment to the paper that is).

That's fine, except then anyone who has a link to that paper would have to tidy their paper. Not such a bad idea
And it could leave certain people who thrive on criticising other theories 'high & dry' too - lol!


What I mentioned above was to clear the paper of all listings. Say if you'd specifically enter the number of the withdrawn paper, you would get the the title/author and the note 'paper withdrawn' - maybe a comment of the author. But the paper does not again appear in the 'new papers' listing under 'replacement', and does no longer show up in search results.

Regarding the existing references to papers: First, it would teach people not to cite papers without having read and understood them, the habit of which I think is a bad development anyhow. Second, I cite very very rarely non-published papers that are on the arxiv only, and if I look at reference lists in other publications I see the same. The possible exception are review articles and lecture notes, things like that. But especially when it comes to experimental papers where I can't tell whether their analysis is waterproof, I would never cite a result that didn't go through peer review. Third, even if a paper contained a reference to a withdrawn paper, that would still tell you a lot if the author argued with that reference...

Best,

B.

Kris Krogh said...

Dear Eric Gisse,

Glad you got a laugh out of that "high-accuracy measurement." A similar paper by the same author, claiming 6% accuracy, was published in Classical and Quantum Gravity last year. This one was a follow-on.

Concerning my own paper , John Wheeler's response was different than yours. Here's a reply to your criticisms, which seem to be based on a very brief reading of the paper:

You've raised the issue of gravitational dipole radiation and PSR 1913+16, without mentioning that those topics are discussed in my Section 12. This theory predicts very little dipole radiation, and effectively the same orbital decay for PSR 1913+16 as general relativity.

(Also see Section 36.1 of Miser, Thorne and Wheeler on gravitational dipole radiation in a theory analogous to electromagnetism. For a system conserving angular momentum, it's zero.)

The assumption of linear superposition of gravitational potentials in this theory is consistent with bodies having significant self-energy. (That is shown explicitly for the lunar orbit in Section 8.) Note that not only times and lengths, but masses also are affected by gravitational potentials here.

The potentials produced by n bodies brought together is not the linear sum of their potentials when they were separate. But the potentials produced by whatever masses result do add linearly. In the derivation of c you referred to (Appendix A), the mass shells are taken to have whatever amounts of matter are needed to produce linear potential steps.

There is also the funny part where you assume SR holds despite SR assuming homogeneity of space and time, which is totally thrown out the window with a gravitational potential.

Here you've raised a valid concern, but may have overlooked this statement in the paper:

Although preferred-frame effects are possible in this theory, it can be shown they are small for existing Solar System experiments. For simplicity, this paper treats the Solar and Earth-Moon systems as being at rest in the preferred frame of the universe. (Corrections for their additional motions will be given in a subsequent paper.)

In this theory, special relativity is only obeyed strictly where gravitational potentials are uniform, as in that derivation of c. That is, when gravitational fields are absent. That's also true for general relativity. (See Chaper 7 of Misner, Thorne and Wheeler, "Incompatibility of Gravity and Special Relativity.")

In a gravitating system with the Solar System's velocity relative to the CMB rest frame, you do get some very small preferred-frame effects in this theory. It turns out they are not at all detectable in the lunar ranging experiment, but there might be other ways to detect them.

Rather than write that up as a separate paper, I'm putting it into a version 24 of the same ArXiv paper. (This is partly how you end up with so many versions. If there is any area I haven't discussed, general relativists like to assume the theory is wrong there.)

Thank you for pointing out the typo's in Dr. Nordtvedt's name. It is spelled correctly 11 times in the paper, but I twice omitted the "d". (At least I got it right in the acknowledgements.) Apologies to Dr. Nordtvedt.

If you have any other concerns, I would definitely like to hear them. But please send them by email, in case Bee and Stefan's readers aren't interested.

Prof Anonym said...

"I cite very very rarely non-published papers that are on the arxiv only"

Perhaps I am misunderstanding you, but you seem to be saying in effect that you never cite any paper that is less than about 6 months old. That seems very extreme indeed! I think that you are placing WAY too much faith in the refereeing process. I, and I think a *lot* of people, find referee reports uniformly worthless, whether positive or negative. Furthermore, there do exist very reputable people who *refuse* to submit papers to journals, as a matter of principle. One rather famous person who does this [he has co-authored with Ed Witten] says: if it is good it will have an impact. If it isn't it won't. Nobody reads journals, so why should I submit? I think you will find this attitude is more common than you realize.

Something to bear in mind: you may think that a paper is wrong, but the author may not agree. He is entitled to his opinion. I recently got an extremely rude email from somebody claiming that there was an error in one of my papers. I wrote back very politely to say why I disagreed. I got an even ruder response; then I put this person in my spam file. I do not welcome such correspondence, and see no need to pay attention to it. Nor have I ever written to anyone to tell him that his paper is wrong.

I think we will have to agree to disagree. Yes, there are bad papers out there citing other bad papers. Yes, there are people who know that their papers are wrong and who refuse to retract. I still think that *boring correct papers* are far far more numerous, and pose a far greater threat to the field, than wrong papers.

John said...

Should people be allowed to withdraw their papers from the arXiv if they were caught having plagiarized them? I don't think so.

In general I'm in favor of having the arXiv be a permanent record. I always submit replacement versions of my papers when I discover mistakes in them. If it means people discover I'm not perfect, so be it!

Lumo said...

Dear Bee,

I clearly agree with your general point, assuming that it is yours, that there exist all kinds of pressures on people to hide changes, hide errors they have found, change how the history actually looked like, and so forth.

Clearly, some people are more likely to act dishonestly and some people are less likely to do so. Someone who tries to design optimal policies for revision etc. must be aware of these phenomena. He or she should care whether people are led to honest behavior, whether the actual contributions will be correctly judged, and whether people will be able to get informed about the actual current status of some statements and theories.

All the best
Lubos

CapitalistImperialistPig said...

Bee,

A lot of good reasons have been mentioned for not allowing papers to be withdrawn. I like this one due to Neils Bohr:
Theoretical physicists have an important advantage over philosophers. We have all written papers that were subsequently proved to be wrong.

Bee said...

Hi Prof. Anonymous:

You have understood me correctly. I cite papers that I used as a prerequisite for a calculation, as well as those who get credits for ideas that were not my own. I usually need more than 6 months to write a paper, so I just don't have citations younger than that. (The possible exception would be if I got scooped on a paper, luckily this has never happened.) I try to avoid citing everybody around there who has possibly worked on something more or less vaguely related (same keyword?), papers that I probably didn't even read. If somebody wants that, he can do a topic search at SPIRES. I occasionally cite papers that are not directly necessary to understand my works just because they are particularly useful, clearly and well written, and I want to draw attention to them.

Besides this however, I admit of being guilty to cite people who keep bothering me with emails to add a reference to their work, just because I want them to shut up. I guess that's not a very good behaviour :-(

Anyway.

Furthermore, there do exist very reputable people who *refuse* to submit papers to journals, as a matter of principle. One rather famous person who does this [he has co-authored with Ed Witten] says: if it is good it will have an impact. If it isn't it won't. Nobody reads journals, so why should I submit? I think you will find this attitude is more common than you realize.

Well, as you said I think we just disagree. I don't know what field you work in, but if you'd maybe look around outside hep-th, and outside your country you'd realize that this attitude is far LESS common than you want to believe. Look at (formerly) nucl-ph/cond-math/astro-ph (i.e. the 'big' fields) etc etc. Almost everybody I know publishes, and those who don't wish they could (restrictions apply to very senior people). Show me the postdoc who got hired with 10 unpublished papers. Ed Witten can afford that, and I guess for him peer review is unnecessary anyhow, but this is hardly very representative for all of theoretical physics. As you say, its the 'reputable' people for whom it's no longer necessary. The question is, how do you get there? What do you think we are coming to if there is only the arxiv and no peer review? Well, let me tell you: papers get rated by peer citations. Much like blogs get judged by number of links from peers. Is that were you want science to go?


Dear Lubos,

I agree. I think I'll have to mark this day in my calendar.

Dear CIP,

Yes, I know. People have raised all kinds of good reasons against. I am still wondering why nobody acknowledges reasons for. I am pretty sure Bohr was referring to published papers, so it doesn't apply to the arxiv.

How about this setup. It is possible to withdraw a paper with the following result: the paper and all previous versions remain on the arxiv, but if you decide to withdraw it does not re-appear in the listing, and does not show up any longer in the usual search. One could take out all the withdrawn papers in a separate database, such that they remain available if one is specifically targeting one, but for the usual recherche one doesn't get bothered with all the 'crap'. So, basically one could shift a paper into the trashcan if one realizes it belongs there, but can't empty the trashcan.

Best,

B.

Bee said...

Hi Prof Anonymous:

Forgot to mention

Something to bear in mind: you may think that a paper is wrong, but the author may not agree. He is entitled to his opinion.

Sure. I am aware that in case of a disagreement it might very well be (in fact is likely) that I am wrong and the author is right. There are plenty of unsettled arguments of all kinds around there. The cases I was referring to were those in which an author agreed on having written crap but refused to admit it publicly. Best,

B.

arivero said...

How about this setup. It is possible to withdraw a paper with the following result: the paper and all previous versions remain on the arxiv, but if you decide to withdraw it does not re-appear in the listing, and does not show up any longer in the usual search.

Yes, that could work. My problem was to avoid editor-required withdrawal.

Cynthia said...

Hi CIP, looks like that quote by Niels Bohr might be the inspiration behind Wolfgang Pauli's "not even wrong"...

Anonymous said...

Concerning the comments by K. Krogh on the "comic" paper(s) by L. Iorio on the test of the Lense-Thirring effect with the Mars Global Surveyor (MGS) in the gravitational field of Mars,
the reply by Iorio to Krogh, is in print on the peer-reviewed journal Central European Journal of Physics (CEJP), after it was already accepted in the past by Journal of Gravitational Physics (JGP) which ceased its activity before publishing that paper.