View Full Version : Subject: Was WMAP wrong?
Garth Barber
Dec3-04, 04:48 PM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE>\nThe WMAP data is angular in nature; conformal transformations preserve =\nangles, therefore the data is not only consistent with a flat universe =\nbut with conformally flat universes.\n\nThe anomalies at the largest angular scales, i.e. the absence of the =\npredicted anisotropies over those scales, may be explained by the =\nuniverse being finite so there was not enough room for these anisotropic =\nfluctuations to form.\n\nPossible conformally flat yet finite universe include: a cylinder, a =\ncone and a torus.\n\nThe lack of these largest fluctuations may therefore be evidence of a =\nnon-standard cosmological topology and the need to modify gravitational =\ntheory.\n\n\n\nThe apparent orientation of these wide-angle anisotropies with the =\necliptic may be evidence of a Machian-type influence of the cosmological =\ndistribution of mass on local physics, and therefore again an indication =\nof a non-standard gravitational theory.\n\n\n\nGarth\n\n\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form"> View this Usenet post in original ASCII form </a></div><P></jabberwocky>The WMAP data is angular in nature; conformal transformations preserve =
angles, therefore the data is not only consistent with a flat universe =
but with conformally flat universes.
The anomalies at the largest angular scales, i.e. the absence of the =
predicted anisotropies over those scales, may be explained by the =
universe being finite so there was not enough room for these anisotropic =
fluctuations to form.
Possible conformally flat yet finite universe include: a cylinder, a =
cone and a torus.
The lack of these largest fluctuations may therefore be evidence of a =
non-standard cosmological topology and the need to modify gravitational =
theory.
The apparent orientation of these wide-angle anisotropies with the =
ecliptic may be evidence of a Machian-type influence of the cosmological =
distribution of mass on local physics, and therefore again an indication =
of a non-standard gravitational theory.
Garth
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE><ebunn@lfa221051.richmond.edu> wrote in message\nnews:col9mm\\$uhf\\$1@lfa222122.richmond. edu...\n> In article <24a23f36.0412010138.399c834f@posting.google.com>, \n> Thomas Larsson <thomas_larsson_01@hotmail.com> wrote:\n> >A second look at the publicly available WMAP data reveals anomalies at\n> >the largest angular scales (> 60=B0). Strong correlation with the\n> >orientation of the solar system (ecliptic plane) and with its motion\n> >(measured as the CMB dipole) showed up, see\n> >\n> >http://www.cerncourier.com/main/article/44/10/4\n> >\n> >Anyone know what the significance of this is?\n>\n> The paper referred to in this article is\n>\n> http://www.arxiv.org/pdf/astro-ph/0403353\n>\n> This is part of a series of results suggesting that there are\n> statistical anomalies in the largest-scale multipoles of the microwave\n> background as measured by WMAP.\n\nHere are some more references pointing at the same (or possibly\nrelated) "problems":\n\nhttp://arxiv.org/abs/astro-ph/0312305\nhttp://arxiv.org/abs/astro-ph/0402399\nhttp://arxiv.org/abs/astro-ph/0402396\nhttp://arxiv.org/abs/astro-ph/0404037\n\nhttp://arxiv.org/abs/astro-ph/0311430\n\n> The original hint of trouble was that there is less overall\n> large-scale power than theoretical models predicted. This was first\n> quantified as a low quadrupole in the COBE data and was confirmed with\n> much better data by WMAP. It turns out not to be just the quadrupole\n> but the octupole as well.\n>\n> The statistical significance of that discrepancy isn\'t all that great,\n> because there\'s a lot of cosmic variance in the first few multipoles.\n> Essentially, that just means that there are very few modes we can\n> sample on those largest scales, so there\'s lots of room for random\n> fluctuations.\n\nYes, that it true. There appears to be no general consensus with\nrespect to the significance of the results. Whereas some papers\nquote deviations from gaussianity confirmed at 99.9% percent\nconfidence level (or so), others say that the data are still within\nexpectations.\n\nWhat I find particularly troubling is not so much the low multipoles\n[which hint at non-negligible curvature], is the apparently well\nconfirmed common alignment of several different axes, derived\nfrom the quadropole and octupole, but also from other observations\npointing to an intrinsic anisotropy of the universe.\n\nTo quote from http://arxiv.org/abs/astro-ph/0311430\n\n"The five coincident axes indicate physical correlation and\nanisotropic properties of the cosmic medium not predicted\nby the conventional Big Bang scenario."\n\nNow this sounds almost like a "smoking gun".\n\nBut:\n\nApparently the directions that can be extracted from the\nquadrupole and octupole data are aligned with the dipole.\nWhat I wonder (but I haven\'t seen it addressed in any\npaper) whether this type of alignment is really so unexpected?\nI am not an expert, but couldn\'t the apparent alignment\nof the quadrupole and octupole planes with the dipole\nnot just be a second and third order effect of the fact, that\nthe sun moves with respect to the (presumably) isotropic\nCMBR? This obviously is the cause of the dipole. But does\nthis motion result in a *pure* dipole or does in result in\nsome dipole with higher order corrections? I guess this\ncan be answered trivially?\n\n> It\'s hard to be sure just how worried we should be about these\n> coincidences. They suffer from the classic problem of assigning\n> statistical significances a posteriori: if you first notice something\n> strange in the data, and then calculate its statistical significance,\n> you\'re going to get a high significance -- after all, the reason you\n> noticed this thing in the first place is that it was unlikely!\n\nThat is always the problem with unexpected effects. On the other\nhand the coincidences appear to be persistent, rather pointing\nto some not yet understood systematics. But as long as the origin\nof the systematics is not known, it is too early to say that inflation\nhas been debunked or the FRW-model is failing.\n\nBest MP\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form"> View this Usenet post in original ASCII form </a></div><P></jabberwocky><ebunn@lfa221051.richmond.edu> wrote in message
news:col9mm$uhf$1@lfa222122.richmond.edu...
> In article <24a23f36.0412010138.399c834f@posting.google.com>,
> Thomas Larsson <thomas_larsson_01@hotmail.com> wrote:
> >A second look at the publicly available WMAP data reveals anomalies at
> >the largest angular scales (> 60=B0). Strong correlation with the
> >orientation of the solar system (ecliptic plane) and with its motion
> >(measured as the CMB dipole) showed up, see
> >
> >http://www.cerncourier.com/main/article/44/10/4
> >
> >Anyone know what the significance of this is?
>
> The paper referred to in this article is
>
> http://www.arxiv.org/pdf/http://www.arxiv.org/abs/astro-ph/0403353
>
> This is part of a series of results suggesting that there are
> statistical anomalies in the largest-scale multipoles of the microwave
> background as measured by WMAP.
Here are some more references pointing at the same (or possibly
related) "problems":
http://arxiv.org/abs/http://www.arxiv.org/abs/astro-ph/0312305
http://arxiv.org/abs/http://www.arxiv.org/abs/astro-ph/0402399
http://arxiv.org/abs/http://www.arxiv.org/abs/astro-ph/0402396
http://arxiv.org/abs/http://www.arxiv.org/abs/astro-ph/0404037
http://arxiv.org/abs/http://www.arxiv.org/abs/astro-ph/0311430
> The original hint of trouble was that there is less overall
> large-scale power than theoretical models predicted. This was first
> quantified as a low quadrupole in the COBE data and was confirmed with
> much better data by WMAP. It turns out not to be just the quadrupole
> but the octupole as well.
>
> The statistical significance of that discrepancy isn't all that great,
> because there's a lot of cosmic variance in the first few multipoles.
> Essentially, that just means that there are very few modes we can
> sample on those largest scales, so there's lots of room for random
> fluctuations.
Yes, that it true. There appears to be no general consensus with
respect to the significance of the results. Whereas some papers
quote deviations from gaussianity confirmed at 99.9% percent
confidence level (or so), others say that the data are still within
expectations.
What I find particularly troubling is not so much the low multipoles
[which hint at non-negligible curvature], is the apparently well
confirmed common alignment of several different axes, derived
from the quadropole and octupole, but also from other observations
pointing to an intrinsic anisotropy of the universe.
To quote from http://arxiv.org/abs/http://www.arxiv.org/abs/astro-ph/0311430
"The five coincident axes indicate physical correlation and
anisotropic properties of the cosmic medium not predicted
by the conventional Big Bang scenario."
Now this sounds almost like a "smoking gun".
But:
Apparently the directions that can be extracted from the
quadrupole and octupole data are aligned with the dipole.
What I wonder (but I haven't seen it addressed in any
paper) whether this type of alignment is really so unexpected?
I am not an expert, but couldn't the apparent alignment
of the quadrupole and octupole planes with the dipole
not just be a second and third order effect of the fact, that
the sun moves with respect to the (presumably) isotropic
CMBR? This obviously is the cause of the dipole. But does
this motion result in a *pure* dipole or does in result in
some dipole with higher order corrections? I guess this
can be answered trivially?
> It's hard to be sure just how worried we should be about these
> coincidences. They suffer from the classic problem of assigning
> statistical significances a posteriori: if you first notice something
> strange in the data, and then calculate its statistical significance,
> you're going to get a high significance -- after all, the reason you
> noticed this thing in the first place is that it was unlikely!
That is always the problem with unexpected effects. On the other
hand the coincidences appear to be persistent, rather pointing
to some not yet understood systematics. But as long as the origin
of the systematics is not known, it is too early to say that inflation
has been debunked or the FRW-model is failing.
Best MP
Danny Ross Lunsford
Dec3-04, 04:52 PM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE>ebunn@lfa221051.richmond.edu wrote in message\n\n> It\'s hard to be sure just how worried we should be about these\n> coincidences. They suffer from the classic problem of assigning\n> statistical significances a posteriori: if you first notice something\n> strange in the data, and then calculate its statistical significance,\n> you\'re going to get a high significance -- after all, the reason you\n> noticed this thing in the first place is that it was unlikely!\n\nThis is what I find worrisome - the HBB needs inflation to survive,\nand the latter has very much a "deus ex machina" Alice-in-Wonderland\nquality. Then we are told - surprise! - that the WMAP data exactly\nmatches the inflationary scenario. Now we have an analysis that casts\ngrave doubt on the first one. What are we supposed to believe? It\nappears from this paper that one can choose between the fairy-tale of\ninflation with yet another epicycle to get it out of this latest jam,\nor simply admit that there are local (likely galactic or\nintergalactic) contributions to the MB and that we should start\nlooking for better theoretical models, starting at the bottom. The\nlatter would have been the obvious choice in the pre-string era of\nsane science.\n\n-drl\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form"> View this Usenet post in original ASCII form </a></div><P></jabberwocky>ebunn@lfa221051.richmond.edu wrote in message
> It's hard to be sure just how worried we should be about these
> coincidences. They suffer from the classic problem of assigning
> statistical significances a posteriori: if you first notice something
> strange in the data, and then calculate its statistical significance,
> you're going to get a high significance -- after all, the reason you
> noticed this thing in the first place is that it was unlikely!
This is what I find worrisome - the HBB needs inflation to survive,
and the latter has very much a "deus ex machina" Alice-in-Wonderland
quality. Then we are told - surprise! - that the WMAP data exactly
matches the inflationary scenario. Now we have an analysis that casts
grave doubt on the first one. What are we supposed to believe? It
appears from this paper that one can choose between the fairy-tale of
inflation with yet another epicycle to get it out of this latest jam,
or simply admit that there are local (likely galactic or
intergalactic) contributions to the MB and that we should start
looking for better theoretical models, starting at the bottom. The
latter would have been the obvious choice in the pre-string era of
sane science.
-drl
g starkman
Dec3-04, 04:52 PM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE>As an author of the study in question, I would like to suggest that\nTed Bunn\'s conclusion should itself be taken with a grain of salt.\nWhile it is true that "a posteori statistics" should be treated\ncautiously, it is also true that when there are enough pieces of a\nposteori evidence stacking up of something strange you need to pay\nattention. So here are the pieces of evidence that something odd is\nhappening with the large angular scale CMB:\n\n(1) There is a lack of power at large angles. (The low notes of the\nuniverse are weaker than they should be.) This has been known for\nabout a decade and confirmed by WMAP. Moreover, they occur in just\nsuch a combination that there is much less correlation between spots\non the sky 60 degrees or more apart than our theories predict. (WMAP\n1 year papers say this is unlikely at the 99.85% level)\n\n(2) The quadrupole and the octopole align. (The quadrupole and the\noctopole are the two lowest sets of "notes".) This was first pointed\nout by de Oliveira Costa and collaborators, but, as pointed out by the\npaper in question, the alignment is stronger than they realized. This\nis because to properly compare, one should first subtract from the\nquadrupole the piece caused by the motion of the solar system through\nthe universe. The allignment is unlikely at the 99.86% level.\nThis alignment is independent of the lack of power.\n\n(3) The quadrupole and the octopole align with the ecliptic plane.\nThis is a main point of the paper in question. The details of how\nthey align are technical, but this is again unlikely at the >99%\nlevel. In addition, they\nare alligned with the ecliptic, again at an unlikely level.\n\n(3b) Alternately, you can view this alignment as being an alignment\nwith the dipole -- the direction of the solar system\'s motion through\nth euniverse. Viewed this way, the allignment is unlikely at the\n>99.5%.\n\n(3c) Related, and unquantified as to its unlikelihood, is what is seen\nin figure 2 of that paper -- the ecliptic threads its way carefully\nthrough the hot and cold spots of the combined quadrupole-octopole\nsky.\n\n(4) Other authors have divided the sky into hemi-spheres and looked at\nthe relative power on one hemisphere and the other. Strangely, they\nfind that this ratio can be unexpectedly large, and even more\nstrangely that the hemispheres with the largest ratio are the ones\ndivided by the ecliptic -- the plane of the solar system. They also\nfind this when they exclude the quadrupole and octopole, making this\neffect independent statistically of (1), (2) and (3).\n\nThere are other such effects that I oculd list, but I think you get\nthe point. While each of these is a posteori, they are also each odd,\nand, we would argue, taken together they are seriously troubling.\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form"> View this Usenet post in original ASCII form </a></div><P></jabberwocky>As an author of the study in question, I would like to suggest that
Ted Bunn's conclusion should itself be taken with a grain of salt.
While it is true that "a posteori statistics" should be treated
cautiously, it is also true that when there are enough pieces of a
posteori evidence stacking up of something strange you need to pay
attention. So here are the pieces of evidence that something odd is
happening with the large angular scale CMB:
(1) There is a lack of power at large angles. (The low notes of the
universe are weaker than they should be.) This has been known for
about a decade and confirmed by WMAP. Moreover, they occur in just
such a combination that there is much less correlation between spots
on the sky 60 degrees or more apart than our theories predict. (WMAP
1 year papers say this is unlikely at the 99.85% level)
(2) The quadrupole and the octopole align. (The quadrupole and the
octopole are the two lowest sets of "notes".) This was first pointed
out by de Oliveira Costa and collaborators, but, as pointed out by the
paper in question, the alignment is stronger than they realized. This
is because to properly compare, one should first subtract from the
quadrupole the piece caused by the motion of the solar system through
the universe. The allignment is unlikely at the 99.86% level.
This alignment is independent of the lack of power.
(3) The quadrupole and the octopole align with the ecliptic plane.
This is a main point of the paper in question. The details of how
they align are technical, but this is again unlikely at the >99%
level. In addition, they
are alligned with the ecliptic, again at an unlikely level.
(3b) Alternately, you can view this alignment as being an alignment
with the dipole -- the direction of the solar system's motion through
th euniverse. Viewed this way, the allignment is unlikely at the
>99.5%.
(3c) Related, and unquantified as to its unlikelihood, is what is seen
in figure 2 of that paper -- the ecliptic threads its way carefully
through the hot and cold spots of the combined quadrupole-octopole
sky.
(4) Other authors have divided the sky into hemi-spheres and looked at
the relative power on one hemisphere and the other. Strangely, they
find that this ratio can be unexpectedly large, and even more
strangely that the hemispheres with the largest ratio are the ones
divided by the ecliptic -- the plane of the solar system. They also
find this when they exclude the quadrupole and octopole, making this
effect independent statistically of (1), (2) and (3).
There are other such effects that I oculd list, but I think you get
the point. While each of these is a posteori, they are also each odd,
and, we would argue, taken together they are seriously troubling.
greywolf42
Dec3-04, 04:53 PM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE><ebunn@lfa221051.richmond.edu> wrote in message\nnews:col9mm\\$uhf\\$1@lfa222122.richmond. edu...\n> In article <24a23f36.0412010138.399c834f@posting.google.com>, \n> Thomas Larsson <thomas_larsson_01@hotmail.com> wrote:\n> >A second look at the publicly available WMAP data reveals anomalies at\n> >the largest angular scales (> 60°). Strong correlation with the\n> >orientation of the solar system (ecliptic plane) and with its motion\n> >(measured as the CMB dipole) showed up, see\n> >\n> >http://www.cerncourier.com/main/article/44/10/4\n> >\n> >Anyone know what the significance of this is?\n>\n> The paper referred to in this article is\n>\n> http://www.arxiv.org/pdf/astro-ph/0403353\n>\n> This is part of a series of results suggesting that there are\n> statistical anomalies in the largest-scale multipoles of the microwave\n> background as measured by WMAP.\n>\n> The original hint of trouble was that there is less overall\n> large-scale power than theoretical models predicted. This was first\n> quantified as a low quadrupole in the COBE data and was confirmed with\n> much better data by WMAP. It turns out not to be just the quadrupole\n> but the octupole as well.\n>\n> The statistical significance of that discrepancy isn\'t all that great,\n> because there\'s a lot of cosmic variance in the first few multipoles.\n\nWell, that is the ad hoc rationalization.\n\n> Essentially, that just means that there are very few modes we can\n> sample on those largest scales, so there\'s lots of room for random\n> fluctuations.\n\nBut this mode isn\'t random at all. Random was the prediction. Instead it\nwas found to be linked directly and significantly to the ecliptic plane.\n\n> What Schwarz et al. have done in this paper is quantify the\n> *directions* of the quadrupole and octupole and point out that there\n> are funny-seeming coincidences.\n\nActually, "coincidence" was not mentioned.\n\n> This is a method of quantifying some\n> more qualitative observations made earlier by Tegmark, de\n> Oliveira-Costa, and collaborators.\n>\n> It\'s hard to be sure just how worried we should be about these\n> coincidences.\n\nThat depends on how worried one gets when observations contradict theory.\n\n> They suffer from the classic problem of assigning\n> statistical significances a posteriori: if you first notice something\n> strange in the data, and then calculate its statistical significance,\n> you\'re going to get a high significance -- after all, the reason you\n> noticed this thing in the first place is that it was unlikely!\n\nThis is a completely unsupported and incorrect statement.\n\n> In this particular case, there are many different "coincidences" that\n> could have arisen between all the different vectors that are lying\n> around. When assigning a statistical significance to the coincidence\n> that was found, one should take into account all of the other\n> coincidences that could have been found but weren\'t. I haven\'t read\n> this paper carefully enough to be 100% sure, but it doesn\'t look to me\n> like they\'ve done that.\n\nPresumably, statistical significance can be properly applied. In this case,\nyou have not looked at the statistics, but are merely claiming that an error\nin methodology has been made.\n\n> I\'m not saying that there\'s nothing to this result -- there may well\n> be. I think it\'s something that people in the community should\n> continue to scrutinize. Although there\'s not, in my opinion, a\n> smoking gun at the moment, there are hints that something interesting\n> may be going on.\n\nWell, if you always start by dismissing anomalous results, you will have a\nhard time seeing smoking guns.\n\n--\ngreywolf42\nubi dubium ibi libertas\n{remove planet for return e-mail}\n\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form"> View this Usenet post in original ASCII form </a></div><P></jabberwocky><ebunn@lfa221051.richmond.edu> wrote in message
news:col9mm$uhf$1@lfa222122.richmond.edu...
> In article <24a23f36.0412010138.399c834f@posting.google.com>,
> Thomas Larsson <thomas_larsson_01@hotmail.com> wrote:
> >A second look at the publicly available WMAP data reveals anomalies at
> >the largest angular scales (> 60°). Strong correlation with the
> >orientation of the solar system (ecliptic plane) and with its motion
> >(measured as the CMB dipole) showed up, see
> >
> >http://www.cerncourier.com/main/article/44/10/4
> >
> >Anyone know what the significance of this is?
>
> The paper referred to in this article is
>
> http://www.arxiv.org/pdf/http://www.arxiv.org/abs/astro-ph/0403353
>
> This is part of a series of results suggesting that there are
> statistical anomalies in the largest-scale multipoles of the microwave
> background as measured by WMAP.
>
> The original hint of trouble was that there is less overall
> large-scale power than theoretical models predicted. This was first
> quantified as a low quadrupole in the COBE data and was confirmed with
> much better data by WMAP. It turns out not to be just the quadrupole
> but the octupole as well.
>
> The statistical significance of that discrepancy isn't all that great,
> because there's a lot of cosmic variance in the first few multipoles.
Well, that is the ad hoc rationalization.
> Essentially, that just means that there are very few modes we can
> sample on those largest scales, so there's lots of room for random
> fluctuations.
But this mode isn't random at all. Random was the prediction. Instead it
was found to be linked directly and significantly to the ecliptic plane.
> What Schwarz et al. have done in this paper is quantify the
> *directions* of the quadrupole and octupole and point out that there
> are funny-seeming coincidences.
Actually, "coincidence" was not mentioned.
> This is a method of quantifying some
> more qualitative observations made earlier by Tegmark, de
> Oliveira-Costa, and collaborators.
>
> It's hard to be sure just how worried we should be about these
> coincidences.
That depends on how worried one gets when observations contradict theory.
> They suffer from the classic problem of assigning
> statistical significances a posteriori: if you first notice something
> strange in the data, and then calculate its statistical significance,
> you're going to get a high significance -- after all, the reason you
> noticed this thing in the first place is that it was unlikely!
This is a completely unsupported and incorrect statement.
> In this particular case, there are many different "coincidences" that
> could have arisen between all the different vectors that are lying
> around. When assigning a statistical significance to the coincidence
> that was found, one should take into account all of the other
> coincidences that could have been found but weren't. I haven't read
> this paper carefully enough to be 100% sure, but it doesn't look to me
> like they've done that.
Presumably, statistical significance can be properly applied. In this case,
you have not looked at the statistics, but are merely claiming that an error
in methodology has been made.
> I'm not saying that there's nothing to this result -- there may well
> be. I think it's something that people in the community should
> continue to scrutinize. Although there's not, in my opinion, a
> smoking gun at the moment, there are hints that something interesting
> may be going on.
Well, if you always start by dismissing anomalous results, you will have a
hard time seeing smoking guns.
--
greywolf42
ubi dubium ibi libertas
{remove planet for return e-mail}
ebunn@lfa221051.richmond.edu
Dec7-04, 08:06 AM
<jabberwocky><div class="vbmenu_control"><a href="jabberwocky:;" onClick="newWindow=window.open('','usenetCode','toolbar=no, location=no,scrollbars=yes,resizable=yes,status=no ,width=650,height=400'); newWindow.document.write('<HTML><HEAD><TITLE>Usenet ASCII</TITLE></HEAD><BODY topmargin=0 leftmargin=0 BGCOLOR=#F1F1F1><table border=0 width=625><td bgcolor=midnightblue><font color=#F1F1F1>This Usenet message\'s original ASCII form: </font></td></tr><tr><td width=449><br><br><font face=courier><UL><PRE>\n\n\nIn article <ce6fcd8e.0412021257.2c164cfa@posting.google.com>, \ng starkman <gds6@po.cwru.edu> wrote:\n>As an author of the study in question, I would like to suggest that\n>Ted Bunn\'s conclusion should itself be taken with a grain of salt.\n>While it is true that "a posteori statistics" should be treated\n>cautiously, it is also true that when there are enough pieces of a\n>posteori evidence stacking up of something strange you need to pay\n>attention.\n\nI just want to thank Glenn Starkman for his clear exposition of the\nsituation. I hope I made it clear to everyone that I don\'t think that\nthese puzzles should be dismissed out of hand. On the contrary, they\nmay be pointing to something interesting. As I tried to say before,\nmy personal feeling is that they don\'t at the moment amount to a\nconclusive case that something funny definitely is going on, but the\npossibilities are highly intriguing. I\'m very glad that people are\nworking hard on trying to figure out what\'s going on here.\n\nIt is quite true that the recently-discovered alignments between\n*directions* of various things are statistically independent of the\nlong-known deficit in *amplitudes* of large-scale modes. The\nfact these are independent does increase the likelihood that something\nfunny is going on.\n\nIt\'s possible that I\'m being too skeptical about this business. I\n"came of age" in the field in the COBE era, when some people in the\nfield were making much too big a deal about the low-quadrupole\n"problem," which was at the time an utter non-problem (due to cosmic\nvariance and a posteriori statistics). I\'m quite willing to admit the\npossibility that that may have made me too hesitant to give credence\nto these more recent developments.\n\nTime will tell!\n\n-Ted\n\n\n--\n[E-mail me at name@domain.edu, as opposed to name@machine.domain.edu.]\n</UL></PRE></font></td></tr></table></BODY><HTML>');"> <IMG SRC=/images/buttons/ip.gif BORDER=0 ALIGN=CENTER ALT="View this Usenet post in original ASCII form"> View this Usenet post in original ASCII form </a></div><P></jabberwocky>In article <ce6fcd8e.0412021257.2c164cfa@posting.google.com>,
g starkman <gds6@po.cwru.edu> wrote:
>As an author of the study in question, I would like to suggest that
>Ted Bunn's conclusion should itself be taken with a grain of salt.
>While it is true that "a posteori statistics" should be treated
>cautiously, it is also true that when there are enough pieces of a
>posteori evidence stacking up of something strange you need to pay
>attention.
I just want to thank Glenn Starkman for his clear exposition of the
situation. I hope I made it clear to everyone that I don't think that
these puzzles should be dismissed out of hand. On the contrary, they
may be pointing to something interesting. As I tried to say before,
my personal feeling is that they don't at the moment amount to a
conclusive case that something funny definitely is going on, but the
possibilities are highly intriguing. I'm very glad that people are
working hard on trying to figure out what's going on here.
It is quite true that the recently-discovered alignments between
*directions* of various things are statistically independent of the
long-known deficit in *amplitudes* of large-scale modes. The
fact these are independent does increase the likelihood that something
funny is going on.
It's possible that I'm being too skeptical about this business. I
"came of age" in the field in the COBE era, when some people in the
field were making much too big a deal about the low-quadrupole
"problem," which was at the time an utter non-problem (due to cosmic
variance and a posteriori statistics). I'm quite willing to admit the
possibility that that may have made me too hesitant to give credence
to these more recent developments.
Time will tell!
-Ted
--
[E-mail me at name@domain.edu, as opposed to name@machine.domain.edu.]
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.