[vsnet-alert 8882] On the OEJV Voting scheme,
or the trap has been sprung
oejvvote at Safe-mail.net
oejvvote at Safe-mail.net
Tue Mar 14 21:40:01 JST 2006
The newly formed OEJV is an interesting and commendable device addressing an important neglected niche in terms of getting information "out there".
However, it's voting system is open to abuse, and such abuse has occurred from some authors. I noted this early on, and as an experiment left a small trap to confirm my suspicions. Something has occurred today not only to confirm fully
these suspicions, further than they had indeed already been confirmed, but also to force me to bring the evidence to light to the wider community.
The OEJV needs mechanisms to protect against this, the simplest of which would be a benign professional or few who could at least give a paper a cursory glance, if of only a few minutes, to at least ascertain it was _science_! Further, amateurs
are not likely sufficiently trained in the remit of scientific publication to be aware of the fine line between what is using a resource for research and what is
The details of the case follow. First a small digression. I have no connection with the author of OEJV 25, the paper that has caused me to address the matter out loud, nor to my knowledge have I had any contact with any SAI employee or member save for a possible email or two cced to Samus around half a dozen years ago, though I am not certain there has even been that much contact.
Recently OEJV 25 was published. It contains details of some red variables brought to light in a little primer I sent to AAVSO DISCUSSION with respect to searching for new Miras in ASAS3 data. The author is one Sokolovsky, K.V. I am referenced in this work, but was completely unaware of it until I saw it and noted the abstract contained my name. Whether it was rejected by PZ(P) or was an experiment in OEJV usage by the SAI team, I know not.
About a week or so ago it received a vote. Two points. It is interesting to note that OEJV papers can have null votes, they don't have to be voted upon.
Very recently it received another vote. The vote is now 3.50, there are two votes total, so the new vote was five points, ie (2+5)/2
And I suspect I know where that vote came from.
Checking all the Nicholson et al papers in OEJV I find that they have all yet again been improved in voting stature by each and every one receiving a ONE point vote within the past
twenty four hours.
For this has been my trap. That is, OEJV 17 is a small piece of pseudoscience pertaining to ostensible "flaming"(sic) Orion variables.
I have used it as the control, the test case. It has received only two sorts of votes. Votes of ONE point, likely from the author(s), and votes of FIVE points, the vast majority of which likely being due to me, but likely not all.
Even before the number of votes were denoted, the unique value of the votes could be used to retrace the voting case.
Today, via an easy indication given by the happenstance good fortune of OEJV 25 existing, I have seen the full evidence. Recently OEJV started including the number of votes, as
well as the voting value. Examining the reference url below you will see which papers have had the most votes.
Well, today those that cared to look will have noticed that all the Nicholson et al papers have very recently received a vote of ONE point. Meanwhile OEJV 25 received a vote of FIVE
points. And that is all the voting that occurred within the past couple of days, as far as I can tell. And I've been keeping checks on it, for that is what you have to do with such an experiment.
And this despite the fact that nearly all the papers involved are allegedly based on the same concept of 2MASS colour utilization for long period variable candidates' prediction. None of the papers present the data well, merely lifting data from, or referring directly to, the source all sky surveys.
Yet while the Nicholson et al papers all received ONE point votes today, Sokolovsky received a FIVE point vote. Again, note that all these votes have occurred in the last 24 hours, with
no others having occured for any other paper that I can see.
I applaud the concept of OEJV. I am impressed that it has put effort into SIMBAD and ADS linkage, instead of living in its own ghetto, as has occured with the re-released PZP and PZ,
which last I checked were still not linked to either, merely hidden away on an little known server. It allows publishing of material IBVS will no longer accept, for although said will
accept informationless papers if photometry was conducted in several bands (a perusal of issues reveals this), it is increasingly rigid in its acceptance protocols.
But this voting system instead of some minimal level of refereeing has failed. Was destined
to, really. And in the case of OEJV has the potential of alienating a group needlessly.
Refereeing isn't the be all and end all it should be, many a professional paper, both old and new, is essentially contentless, neither furthering nor confirming issues, and indeed at times you can find a serious error of basic science that makes one wonder if the referee actually bothered to read it!
Yet the voting system fails.
I would request that Lubos Brat removes ALL votes of FIVE points on Nicholson et al papers _that_ originated from the UK _only_, as they were from me. The especial case is for OEJV 17. I did do a round of five point voting today to balance out the new one point votings that have cropped up, but to the best of my memory I've only every voted on the other papers in question once or twice.
This should leave that paper with about half a dozen votes, if that, primarily votes of ONE from the authors, although it wouldn't surprise me if a FIVE or few had slipped in from others that are incredulous. This is not a paper that would receive a ONE point vote from anyone objective. I rather like the polite comment Ondrej has posted for this one, trying to suggest delicately that data should be checked for its error nature before conclusions can be drawn upon it, a valid comment as yet unreplied to I note.
The case should be highlighted well by this paper.
[Actually, a theory is properly scientific if it can make testable predictions. Therefore I predict all votes of ONE on Nicholson et al papers conducted recently not only stemmed from
the same url, but also were from the same url that voted FIVE for OEJV 25 recently].
If refereeing is intangible, then coding will have to fix it. Clicking on a vote should give a list of all votes. A run of FIVES and ONES from a lookup table should reveal the case to
a view... ...ie not so much vote as battle. Even better, the table should contain the referring (numeric) IP adjacent to each vote...
Removal of all UK five point votes on these papers should fix what harm my experiment may have done. Hopefully it'll help to stop this in the future, but if it is felt I've just been a damn
nuisance, well, such is suchness. Despite having given up on all of variable star astronomy and contacts, except for this one outstanding case (because the experiment was still ongoing), I
would like OEJV to work. The idea and sentiment is right, and I've read enough half baked and empty professional papers not to be put off if only even one paper in ten carries any information.
Information has to be searched for, and searched hard. It don't jump up and down in front of one shouting "cooey!".
VSX will be a database, it will not be able to do what OEJV does, and PZP and PZ are not connected to the world, and IBVS is going more main stream in its acceptance criteria. OEJV only needs to try that bit harder to protect itself and be useful.
Ref : http://var.astro.cz/oejv/oejv.php?lang=en
And that concludes all outstanding exercises and experiments I was involved with, so truly I disappear. Don't all cheer at once.
I've got bored with being mr smartarse, so either look after yourselves, or don't bother. Whichever. But do think things through, stand outside of your niche and try to get an objective overview, and think on to whether it is all meaningful and useful twenty to thirty years down the line when some future researcher is trying to make sense of all this info that gets disseminated. Do some research, learn what is what.
And ever so everso finally, a very little bit of astrophysics goes a very long way. Learn some.
More information about the vsnet-alert