The rumors about the Higgs particle at the Large Hadron Collider [LHC] have begun again, and since that’s all anyone is going to want to talk about until we actually get the news for real, at the ICHEP conference in Melbourne in a couple of weeks, we may as well get started.
[This is especially true since we learned last year that some well-known non-particle-physicist bloggers have information pipelines directly into the experiments. It is perhaps inevitable that there are scientists who see it in their best interest to subvert the scientific process.]
The current hot rumor is that the LHC experiments ATLAS and CMS have seen, in the new 2012 data, very roughly what they saw last December in the 2011 data, at least as far as the signal from a Higgs decaying to two photons (particles of light) in the mass range of 125 GeV/c2. Note I am just repeating what I have seen on other blogs; as a matter of policy, I do not report secret information on this website, and I make no comment on the validity of any rumors.
Suppose this were true; what would it mean? Well, it’s pointless to try to speculate about exactly how many standard deviations you would obtain by combining data in various ways — if I tried, the only thing I could guarantee you was that I’d get it wrong. So rather than kick arbitrary numbers around based on guesswork, let’s talk about the larger picture.
There are two ways to think about the 2011 and 2012 data. One is to think about them as data sets you should combine together to extract the maximum information. That’s what you would ideally do if you had access to all of the data and knew all the subtleties about how to do this. It is not going to be that easy. Among many issues, one is that the data are taken at different energies, 8 TeV per proton-proton collision this year versus 7 TeV last year. So in combining them you are making a theoretical assumption about how the production rate for Higgs particles changes as you change the energy from 7 TeV to 8 TeV. That’s something you may or may not want to do, and at best it makes things rather complicated when you try to interpret the result.
Another way to think about it, as I discussed last week — and I think this will turn out to be more useful as a rule of thumb — is that last year’s data was good for excluding the Standard Model Higgs particle over much of the range for its mass that was available before 2011. From a range of 115 to 800 GeV/c2 we’re now down, after 2011, to a range of something like 120 to 128 (where we can argue about the edges, but it doesn’t matter very much) or above 600. The hints we saw last year at around 125 GeV/c2 (see also here, here and here) were very interesting and suggestive, but weren’t convincing, in part because with such a large range of possibilities to explore, the possibility of such hints appearing by random chance was not that low. That’s due to the so-called “look-elsewhere effect”. One could only obtain high significance by combining lots of different low-significance measurements together, a technique which can be very problematic, especially when some of these measurements are very difficult.
Now we come to 2012. The LHC experiments ATLAS and CMS have just about the same number of collisions so far in 2012 as they had in all of 2011; and since the energy is a bit higher, the production rate for Higgs particles should be a bit higher (by several tens of percent.) So what we’re going to see in the current 2012 data is a bit like a do-over of 2011… except that now the range of possible Higgs masses that needs to be explored is much smaller. The smaller range means that if the same hints were seen in the same place as they were seen in 2011, the probability of this happening in the search region by random chance would be much lower than in 2011. More precisely, this means the difference between the naive significance of any excess in the data, and the significance after accounting for the look-elsewhere effect, will not be very important. Look-elsewhere was a big effect in 2011; but after the 2011 results, it will not be a big effect in 2012. Thus (roughly) if you view the 2011 data as needed to narrow down the search region to a small window, you can roughly take the significance of any reported 2012 excess almost at face value.
So suppose we did see hints in the 2012 data that roughly resemble those in the 2011 data? Then I think you’ll see most everyone agree that the evidence for something real in the data is pretty strong. Of course we’ll need to see this in both the ATLAS and CMS experiments to have full confidence, but let’s assume that’s the case. What would be next?
- First, is what’s being seen in the data a Higgs particle at all? Could it be something else?
- Second, if it is a Higgs particle, is it the simplest possible type of Higgs particle — the Standard Model Higgs particle — or is it something more complicated?
I’ve described in some detail previously the strategy used to answer these questions, so I refer you to that article.
Generally, I suspect that the July 2012 data won’t be enough (even if you did try to combine it with the 2011 data) to shed clear light on these questions. But the strong theoretical prejudice on the first question will clearly be that “what’s being seen is a Higgs particle of some type”; after all, the theory of the weak nuclear force requires there be something Higgs-like somewhere, and the Higgs in this mass range was predicted to show up first as a bump in the two-photon search. I don’t think you’ll find many people who will think it likely that it’s not a Higgs particle, though due diligence will of course be needed to make sure.
The theoretical prejudice on the second question is much weaker, however; we really have no idea whether the Higgs is of Standard Model type or not. But it seems likely we’ll need the full year’s data set before we start making much progress on this crucial issue (though surprises are possible, if the Higgs is sufficiently different from a Standard Model-type Higgs). And indeed, even if the Higgs looks somewhat Standard-Model-Higgs-like by the end of the year, our knowledge will still be vague; through Phase 2 of the Higgs search, we’ll be continuing to address the issue for the rest of the decade, making more and more precise measurements of the properties of the Higgs, looking for any deviation from the Standard Model’s predictions.
All of this assumes the rumors are correct. IF it is true that ATLAS and CMS see something of roughly the same size as last year, in exactly the same place in their plots, then that would mean that evidence for the existence of some type of Higgs particle in the 125 GeV/c2 range had firmed up considerably. And that would be Very Big News. Well — we’ll find out soon enough.
71 Responses
If you want to obtain a great deal from this article then you
have to apply such techniques to your won blog.
I see Peter Woit is lying about me again. I accused certain scientists of risking the analyses by feeding him information, and instead he arrogantly assumes that he himself was the target of the accusation. I specifically referred to “Scientists”, and so it is obvious I didn’t refer to Woit himself — he hasn’t done any useful science in decades.
I’m afraid I agree with Fred and some others
http://www.theatlanticwire.com/global/2012/07/best-physics-gossips-you-should-be-reading/54123/
I agree with Richard (June 18):
“The only people who really care about the results already know, via their own professional rumour mills, pretty much what is going on.
The only other people in the world who care are people like me: amateur science enthusiasts, ..”
I hardly can imagine how any rumour could influence the work of a scientist – as more so, as rumours often arrive in company with their opposite.
As a layman, I would like to hear more of how knowledge is produced by cooperation of large groups of scientists. No grownup will perhaps assume that to be a frictionless process. So I don’t mind the occasional ‘rumour’ about faster than light neutrinos or Einstein proved wrong. We amateurs are grateful for any crumb from the table of the sages an example of which i consider this website, too.
So basically you are screaming at woit for “subverting the scientific process” for reasons that are so secret you cannot explain them here. Wow, there are a lot of wild blogs out there, but this one is taking it to another level.
Prof. Strassler is right,
… “subverting the scientific process” is just a polite unterstatement of PWs activities … 😉
“This is especially true since we learned last year that some well-known non-particle-physicist bloggers have information pipelines directly into the experiments. It is perhaps inevitable that there are scientists who see it in their best interest to subvert the scientific process.”
Yep, obsevring the behaviour and attitudes of the “non-particle-physicist blogger” and his pipeline(s) into the CMS (and other) experiment(s) for example convinced me a long time ago that they intentionally do what they do NOT for science; on the contrary … 🙁
As a consequence, I stopped clicking both sites … 😉
You’re obviously someone who thinks he’s a highly principled individual. I note that you’re still a regular on the blog run by Lubos Motl who is playing his part in spreading rumours on the Higgs.
Gentlemen, gentlemen, perhaps instead of silly arguments you should start doing some serious physics 🙂
I am reminded what Nima Arkani-Hamed said recently – maybe we are stupid, maybe the generations before us were much smarter. Will not elaborate….
“The theoretical prejudice on the second question is much weaker, however; we really have no idea whether the Higgs is of Standard Model type or not. But it seems likely we’ll need the full year’s data set before we start making much progress on this crucial issue (though surprises are possible, if the Higgs is sufficiently different from a Standard Model-type Higgs).”
In what respects might another year of data distinguish a 125 GeV +/- particle that is merely impersonating a Standard Model-type Higgs from the real McCoy?
Presumably we would know, almost immediately, given the nature of the searches that are showing some kind of signal at that mass range already that: (1) this particle has no electromagnetic charge, (2) that this particle experiences weak force decays that include the predicted channels from the Standard Model, (3) this particle has an even integer spin (which follows from diphoton production), (4) this particle lacks QCD color charge (this follows from both diphoton productioon and some of the other decay chains attributed to the Higgs boson signal, and (5) that the mass of this particle is not inconsistent with constraints on the mass of a Standard Model Higgs boson derived from electroweak data, vacuum stability considerations, or unitarity considerations.
One can imagine that an imposter Higgs boson would have different couplings to some fermions than the Standard Model predicts, I suppose. But, if this were the case, would another year of LHC data really reveal that? I support one could imagine a Higgs boson that interacts equally with both right handed and left handed fermions in a way that resembles weak force interactions, even though W and Z bosons only appear to interact with left handed fermions (potentially providing a bridge to a sterile fermion sector), but again, would another year of LHC data reveal that?
I guess I could imagine CP violation with the Higgs boson that differs from the Standard Model prediction value, but even then, a change of a parameter like that, unless it was extremely striking, would seem on a par with the addition to the Standard Model of mass for neutrinos. Nobody talks about neutrino oscillation which supports the existence of neutrino mass as evidence of “non-Standard Model neutrinos.”
The only result I can imagine that would be consistent with data so far, that would be revealed by another year of LHC data but inconsistent with a SM Higgs, would be non-SM branching fractions in its weak force decays.
Am I missing any interesting possibilities?
In my opinion, a spin 2 neutral resonance, a fermion condensate or a Weyl boson (enforcing local scale invariance of the theory) may be listed as “would-be” candidates.
As you say, if there is a particle decaying to two photons it is obviously electrically neutral, spin 0 or 2, color neutral, and consistent with a Higgs.
The statement that it experiences weak force decays (i.e. decays to W’s and Z’s) is still an experimental question; if we observe those things, the next question is whether the ratios of photon decays to W and Z decays are as expected. A failure of this ratio to match the Standard Model is the first clean place the Standard Model higgs hypothesis might break down. The second place is that the rate for decays to photons times the overall production rate (the sheer number of Higgs particles decaying to two photons) might be too large or too small compared to expectations; this too would represent a breakdown. Both of these possibilities are accessible THIS YEAR, if the deviation from the Standard Model is of order 1 (which often happens in models.)
A third possibility is the ratio of production modes of the Higgs might be off. There are several, and their ratio is predicted in the Standard Model. These ratios will be, to some degree, measured this year, and again, large effects that are often seen in models could already be detected in 2012.
Another thing that would be accessible this year would be decays of the Higgs that are forbidden in the Standard Model, or production modes of the Higgs that are forbidden in the Standard Model. Either of these could be accessible with this year’s data.
As for fermion decays, if you muck with the fermion decays you’ll usually change the branching ratios to W’s, Z’s and photons too, so probably we’d see an effect there first.
[A small error in your comment: all Higgs bosons interact with left- and right-handed fermions; that’s necessary for any Higgs to give the fermions mass in the first place, and also required more generally for any spin 0 particle.]
Finally, of course, there could be more than one Higgs particle. Discovery of a charged Higgs particle, or of a second neutral Higgs particle, would of course rule out the Standard Model.
I’m not sure this is the complete list; but it will definitely keep us busy during the long shut-down that begins at the end of the year.
It is troubling that you have to explain in detail to a lecturer at Columbia why spreading rumors is unethical and counterproductive.
I think, like the WikiLeaks people and certain types of investigative journalists, individuals such as Peter do really believe they provide a valuable service. And they often do think about it carefully (clearly Peter is not unthinking.) I am not sure what it would take to convince him that he runs the risk of doing more harm than good. I do think that unless a large group of the experimental community signs a letter asking him and other bloggers to stop this activity, he won’t. He’s certainly not going to listen to me — and why should he? I’m not in the collaborations either; and despite working closely with experimentalists, I have little more visible authority here than he does.
Oh, and you noticed I’m sure, he never did answer my question.
I did answer your question, although it seems not explicitly enough. So, again: I actually know enough about what these experiments are doing to have a good idea of what the consequences of my actions are.
You seem to have a different idea, that there is some top-secret aspect to HEP data analysis that I don’t know about and am ruining because of this ignorance. I don’t believe this, for many reasons. A big one is that I’m sure that if it were true, I’d hear about it quickly through one of many possible channels. For instance, I have friends who work on these experiments who would be sure to explain it to me, as well as many, many blog readers who are quite devoted to e-mailing me and explaining to me things I don’t seem to know and should. In addition, my experience with HEP physicists I don’t know is that they’re not shy people, quite willing to give me a piece of their mind if they think I’m missing an important point. You conjecture that they need to keep this top-secret stuff top-secret, so none of them can explain it to me, but I’ll point out that their record of keeping top-secret stuff top-secret is not very good. Your concept of large number of experimentalists writing a letter to me and other bloggers saying don’t repeat information, but we can’t tell you why you shouldn’t do this is rather absurd. All it would take is one person with a convincing case.
At this late stage in the Higgs discovery game, the big problem caused by information leaking out early on blogs is not a scientific one, but a PR one. I understand that the existence of such information on blogs is a big problem for those tasked with how to communicate with the press and public. It’s a new problem, but not completely different than that of years ago, when information would leak to journalists, who would then
call up CERN and ask for confirmation, sometimes writing news stories about such leaks. CERN is doing now the same sensible thing they did then, saying “no comment, don’t believe anything you hear until our public announcement”. Last year’s experience I think showed that blogger’s leaking information before public announcements is just such a PR problem, and one that they can readily deal with.
Thanks, Peter, for answering the question and making your position clear. That is helpful.
While awaiting your reply I got another private message from an experimentalist indicating disagreement with your view of ‘unblinding’ in the LHC context.
You say LHC experimentalists aren’t shy. I find the reverse. I wish they would say in public what they say to me in private. Then I could stop arguing with you. I can’t blame you for not taking me, a theorist, that seriously on matters like this.
Matt,
Please for me explain to your shy experimentalist contact that I’m well aware that I have no idea how they blind and unblind their data, and that I am sure there are complexities to this that I know nothing about. As I keep pointing out, I have a fairly clear idea of what I don’t know. My reference in comments here to this blinding process was in no way intended to accurately describe what they are doing, just to refer to the general idea. One of the wonderful things I learned long ago about blog comment discussions is that when you’re writing quickly and trying to make a certain point, you leave yourself open to people jumping on anything not quite complete or accurate that you write, or even something accurate that can be misinterpreted. Blog discussions with Jacques Distler were always a joy in this regard…
Again, the issue of blinding is irrelevant to claims that I am ruining their work, since the information on my blog is about unblinded data, what was seen after “opening the box”. If you or your contact can explain how hearing that experiment X got the expected result when they “opened the box” on their gamma-gamma analysis is going to ruin experiment Y’s blinded gamma-gamma analysis, I’m all ears.
It is refreshing to hear a voice of reason from the ATLAS spokesperson, see below:
http://www.nytimes.com/2012/06/20/science/new-data-on-higgs-boson-is-shrouded-in-secrecy-at-cern.html?_r=1
Sadly, there are far too few voices of reason in the modern world.
Very interesting. Thanks for shedding light on the latest rumours.
Regarding theoretical assumptions: Conceptually, there is some model that is used to extract signal from the collision data, and that model is assumed to work over a range of momenta of the partons even at a fixed 7 TeV collision energy. All that changes in going from 7 to 8 TeV is that the momentum distribution of the partons shifts, and the rare events at higher momentum at 7 TeV become less rare at 8 TeV. There is also some small additional range of momentums which were inaccessible at 7 TeV.
Now it is possible that the model that is used to extract signal from the collision data has some free parameters that are recalibrated at each run, based on data obtained from the collisions, at 7 TeV and 8 TeV. It is only if the recalibration shows some unexpected differences between 7 TeV and 8 TeV will there be some unjustified theoretical assumption in combining the results.
No, I am afraid this is not correct. You’re making a mistake.
The Higgs is produced in gluon-gluon collisions, in quark-antiquark collisions, and in quark-quark collisions. The balance between these three production modes is predicted in the Standard Model, but not known ab initio. Your statement does not account for this. The relative balance in Higgs production between these partonic initial states changes as you go from 7 to 8 TeV production, and so the Higgs cross-section (and also the experimental efficiency for detecting Higgs bosons in different final states) changes in a way that is not predicted by knowing the parton distribution functions alone. You must know the relative size of the the various production channels, and that is model-dependent.
The effect is relatively small; don’t get me wrong. But one of the things you want to test is whether the Standard Model for Higgs production is correct. Assuming the Standard Model production modes are the ones in nature does bias your results — slightly, but maybe more than you would like.
Conceptually how would one measure LHC collision results in a phenomenological way independent of Standard Model or other competing SUSY, technicolor models. The parton momentum distributions have to assumed. The unknown function is (using Dorigo’s formulation) σ_point(x_i x_j s, H). Or rather three such functions for each type of collision pair. The Standard Model has a definite prediction for the σ_point, to be sure. But whatever σ_point is, it continues to be the same function at 7 TeV and 8 TeV, it is simply a somewhat different range of x_i, x_j that gets explored.
Anyway, probably no point in arguing.
The point is elementary, Arun. And there is every point in arguing, when there is a simple physics point, and you are misleading other readers.
1) there is a process g g –> H . This has one cross-section.
2) there is a process q q –> q q H. This has a second cross-section.
3) there is a process q qbar –> W/Z H. This has a third cross-section.
The weights between these are different in different models.
For any one of these, you are right of course, the 7 and 8 TeV measurements are easily related. You don’t have to explain to me how parton luminosities work.
But the total cross-sections at 7 and 8 TeV depend on the relative weights of these three processes.
Therefore the measurements of the Higgs signal at 7 and 8 TeV allow one to determine (in part) the relative weights, without any assumption about the underlying model.
That’s how you do it “independent of the Standard Model and other competing SUSY, technicolor models.”
“The Higgs field is a triumph of quantum field theory, using the most advanced theories of our time and building on the successes of decades of research. But then the Higgs field has similar properties to the aether, it’s a perfect, isotropic field filling the universe. It interacts with everything, but it’s surprisingly difficult to see, no matter how hard we look. If we don’t see the Higgs boson, but instead see something else then we could be in for another glorious revolution in physics, and the quantum field theories could seem like quaint approximations of a bygone era.” — Aidan Randle-Conde
http://www.quantumdiaries.org/2011/09/08/higgs-skeptic Higgs Skeptic – Quantum Diaries, Sept. 8, 2011
Is it fair to say that, in theoretical terms, the Higgs field is “a perfect, isotropic field filling the universe”?
Reblogged this on In the Dark and commented:
Reaction to rumours about the Higgs, and a not-entirely good-tempered comment thread about blogging ethics – all in a day’s work for particle physicists, I guess!
Well, I’m sorry I have to say that I agree with Matt in content and form.
Content : his analysis about the 2012 data makes a lot of sense to me. He just says that combining data is not as easy as it may seem. Not that it will never be done. And that we may get a lot of information without it.
Form : I understand his comment on “subverting the scientific process”, which may be somewhat severe, as follows. We are all passionate people, but we have to make a difference between speed and haste. The problem is not that some people withhold information, it is to give to the whole set a consistent interpretation before disclosing it. Disclosing a partial information too soon can be counterproductive. I don’t think scientific information works as public tabloïd-type media information. The name of the game is not to disclose instantly every bit of information. The subtleties we are all dealing with need some sort of digestion which needs time. A stable feedback loop needs a time constant (not too high though to be stable, I agree). The scientific process is to get a consensus. This can be done only if we exchange clear ideas, not necessarily raw data. Am I wrong in interpreting your comment this way ?
I’m sometimes surprised by the prejudice which consists to think there is an ongoing battle between a group of evel people who keep the information secret and the good ones who want to disclose everything at more than lightspeed.
Sometimes, the `tune the cuts to please the higher-ups’ replaces the LEE.
“”So in combining them you are making a theoretical assumption about how the production rate for Higgs particles changes as you change the energy from 7 TeV to 8 TeV. ”
Actually, at 7 TeV or 5 TeV or 8 TeV, the constituents of the protons are what are colliding, and these constituents have some energy distribution with non-trivial spread, and the physicists are making a theoretical assumption that they know the production rate along this energy spectrum. So I’m not sure what great additional assumption we are making in combining 7 TeV and 8 TeV results.
Tommaso Dorigo answered the question
http://www.math.columbia.edu/~woit/wordpress/?p=4772&cpage=1#comment-114354
[Good for Dorigo; you may want a second opinion.]
Let’s not make too big a deal out of this; I said only that you “may or may not want” to combine 7 and 8 TeV results, but I did not say you “should not” combine them.
What is the theoretical assumption that goes into combining 7 and 8 TeV data? You must assume that the Standard Model (or some other model) is correct, so that you know the ratio of the production rates for the Higgs at 8 TeV to the rates at 7 TeV. The various production rates change by different amounts as you go from 7 to 8 TeV, so if you assume you know those ratios, you are assuming something about the world that you actually will want to measure using this data. Of course you may choose to show both uncombined and combined results; the first way would be assumption-free and model-independent; the second would be model-dependent, but in a known way. I prefer (at least initially) to see model-independent results (this is part of my data-interpretation conservatism, which Dorigo generally disagrees with; he prefers to assume the Standard Model is correct until the data shows the assumption is wrong.) So I think it would be good if they do both, if time permits, when they publish their results.
But there won’t be any time for a combination by ICHEP, so it is moot for now.
Replying to Matt below, rather than to Arun above (for some reason I don’t see the “reply” button attached to Matt’s answer).
I do not know where you got that piece of information about what I prefer, but you happen to be correct. It’s a matter of choice, indeed.
Cheers,
T.
Good, glad that we are in agreement on that point, Tommaso. I guessed it from your “Firm Evidence” post.
Matt,
Your commentary is the most honest and objective I have seen so far. For anybody to assume a Higgs from the data at this point seems rather ridiculous as we know many particles that can decay into two photons (why wouldn’t the unknown ones also do this depending upon the mechanism?). The W and Z products are necessary for Higgs confirmation and I don’t see too many discussions about it (except yours). Instead, we see a lot of discussion about the need for the Higgs and the excitement that it is there – this is not objective analysis of scientific data but wishful thinking IMHO.
Well.. I must admit I don’t really understand you at all. I assumed you were just being careful about what to say and whatnot to say, but I’m starting to believe there’s something seriously wrong here.
First of all, how is disclosing information “subverting the scientific process”? What absurd definition of “scientific process” are you using? Some scientists have decided to withhold information, therefore disclosing it goes against the scientific process?
Second, how can you use one data set (2012) for significance while using another (2011) only for limiting the look elsewhere effect? If you’re not going to use the data for what it is, both excluding masses at some significance and excluding the SM background at some significance, couldn’t you arbitrarily choose to use 2012 for the LEE and 2011 for the signal, or both for LEE only, both for signal? The current “scientific process” here is to use the local rejection of the SM background only, ignoring LEE completely and using a high standard of five sigma to make sure there are no false positives claimed as discoveries.
I noticed you left out any actual numbers from the rumors. A “scientific process” would be to take those numbers, combine them in the way previous numbers have been combined (assuming, of course previous combinations predicted official results well enough) and seeing what kind of numbers you get. From this you can estimate how much margin of error you have and make a good guess at how likely this is to give us a discovery.
Obviously, you can follow the “scientific process” and ignore any numbers or mathematics, claiming it’s impossible to know anything, ever.
(Apologies for the bitter tone / sarcasm; I just think you’re going completely overboard with how little info a naive combination would give)
The problem in this case is the data analyses are not finished let alone internally reviewed by those working on them. There’s no question of withholding information, ATLAS and CMS have only just finished collecting data for ICHEP and valid interpretation of those data cannot be instant. There are many people working very hard at the moment to ensure the integrity of results. Those who choose to leak unreviewed results knowing their favourite blogger will disclose them (not Matt although this post does fan the flames a bit), are potentially undermining the work of their colleagues. We saw with supposed faster than light neutrinos from OPERA how rushing to make results public because of rumours can lead to flawed analysis.
It’s also fair to say that the strategy for presentation of results in terms of LEE and/or comibination of data has not necessarily been decided upon by the LHC collaborations. This will happen in the coming weeks; we can’t even say whether ATLAS and CMS will follow the same strategy.
At the moment I think it’s fair to speculate what might be expected from 6/fb of 8TeV data, the size of the dataset is very much public; but making that speculation with unreviewed and far from final numbers adds nothing worthwhile. Anyway the ICHEP Higgs presentations are less than three weeks away, so it’s not that long to wait for the official results!
Mark,
The claim that rumors on blogs had anything to do with OPERA’s decision to go public with bad results is nonsense. See
http://blog.vixra.org/2012/04/20/opera-blames-the-blogosphere/
And in the case of the Higgs, do you really think it’s my or anyone else’s blogging that is driving the schedule for a public announcement? If ATLAS/CMS prematurely release bad results at ICHEP, not my fault….
Peter —
No one is suggesting (or should suggest) that the schedule for the ICHEP results is driven by your blogging (or anyone else’s.) Other forces are at work. That said, you’ve made the decision to release some results that (as I assume was obvious to you) can’t possibly be in final form yet… and that is, to a degree, your fault. [You do realize that the results released last December in rumors also were not yet complete, yes?]
But the real fault lies with whoever sent you those results. I won’t state my opinion of such people in print.
Let me be clear, lest I be misunderstood: I am personally convinced that you are doing what you believe to be right, honest, ethical, etc. I am sure that you are convinced that what you are doing serves science and that in reporting rumors you are doing no damage to the scientific process. It’s true I don’t agree with you, based on all sorts of things I can’t talk about… which doesn’t help the situation, since it’s not helpful to my argument, and is not fair to you. But even if I could talk about those things, unfortunately neither your point of view nor mine is provable; I can’t think of any experiment that we can do to determine who is right. So we may just have to continue to disagree, for now. A few relevant points were made by Torbjorn; maybe others will chime in, so you can see more of how people really feel about it.
I realize it would be impossible for me to convince you that releasing this preliminary information about the Higgs search (or other analyses) actually might be, in the end, damaging to CERN, or to ATLAS and CMS, or worst of all, to the scientific knowledge that we obtain through the Higgs search (or other analyses). But maybe others could do so? Is your policy about releasing preliminary rumors something that you would potentially reconsider? Or do you view it as fixed in stone?
Replying to Peter Woit:
You are putting a lip gloss on this, of course. The issue that you avoid by Looking Elsewhere is the point of undermining.
If the official release says something else than the rumors, there will be an increased workload on scientists. They have to expend effort to inform and defend science more, since some in the public will hear the rumors and maybe exclusively so. Thus it undermines.
Does your rumormongering contribute to the noise? Yes, obviously so since the heading is “The Higgs Discovery” and the original article is “seeing … signal” so implying detection. And the update strengthens this by “announce discovery … or … wait for some overhwhelming convincing standard” and “there is something there”. If there is no detection, and I believe the SNR is expected to fall short of it, there will be undermining since there was overreach. As Strassler says, there will be strong prejudice – but no knowledge according to previously agreed standards, due diligence is missing.
Unless there is any science suggesting that rumors, early speculation and so on contributes to the process of informing the public on science, it is fair from the above observations to conclude that there is undermining, by outsiders none the less.
[Woit is not even wr… a publishing physicist I believe, though it is rumored he sometimes lectures on the subject. In any case, he claims to not be a particle physicist in the rumor post.]
We all want to see the science I’m sure, but we have different opinions and perhaps knowledge on how to go about it. We should respect that there can be problems with all aspects of it!
Peter,
Sure the timeline is set by the conference cycle but your blog was discussing unreviewed data and not what ATLAS and CMS will ultimately present. ICHEP results on 2012 data (Higgs or otherwise) will be preliminary and likely treated with caution until time allows for further analysis to varify any initial suggestions if they’re there. The reviews of ICHEP data are on going, in fact we haven’t even looked at all the data yet – there was an additional 1.3/fb in the last week alone which could easily change the current picture! The decision of what to show or not at ICHEP will be based on the results of the scientific reviews within the collaborations not on what your source(s) feel the first half of the data may or may not be showing. Incidentally I appreciate you do help generate interest in the results, my annoyance is directed more at those within both collaborations who feel it’s ok to circumvent the internal review and give you potentially misleading or incomplete information.
Matt and Mark,
I understand very well that what I write on my blog creates an uncomfortable situation for people at CERN and those working on the experiments. I don’t however see a case that it damages the science or the public understanding of it, and could easily argue the opposite: at a time when inaccurate characterizations of soon to be released results may start flying around, an accurate characterization is helpful. I’m well aware of how early these results are, hope I make that clear in what I write on the blog.
I don’t have a policy of putting on the blog everything I learn, in each case I do think about whether it’s a good idea or not. At another extreme, I’ll point out that I never wrote about the OPERA results on my blog, even after the public announcement, other than to respond to commenters who wanted to discuss OPERA by telling them that it was essentially certainly a mistake, and a really bad idea to say anything about it to the public other than that it should be ignored.
I just checked and there is some inaccuracy in my previous comment. In response to comments I did give a short explanation about why the OPERA result was not to be believed, as part of a multi-part posting mostly about other things.
The great majority of the comments submitted to my blog wanting to discuss OPERA were simply ruthlessly deleted.
Well, about OPERA, your choice is one that I understand. I made the choice that it was better for science that I be a source of accurate information rather than say nothing about it — a teaching moment, as it were. Both choices seem reasonable.
That is beside the point in any case. At no point did either of us release information from OPERA that was not yet made public elsewhere. The blogger who has something to answer for is the one who made the information public before it was final and fully approved by the scientific collaboration. And that was neither you nor me.
This time, however, you are the blogger who has to answer. Regarding your earlier comment, I don’t see how you can or should assume the right to “make things uncomfortable” for other scientists. What have these people done to deserve you making things uncomfortable for them while they’re still trying to get the science done? Afterward, sure, once the presentation is made, do what you want. But right now they’re still hard at work; leave them alone. Who gave you the right to make their lives difficult?
It seems so contrary to your “Not Even Wrong” philosophy that I really don’t get it. What is more important in science than pristine data analysis?
Torbjörn Wrote: If the official release says something else than the rumors, there will be an increased workload on scientists. They have to expend effort to inform and defend science more, since some in the public will hear the rumors and maybe exclusively so. Thus it undermines.
Reply: I think the big problem that Atlas and CMS will have if Peter’s rumors are not true is in defending the 2011 results where there are 3 sigma results in both Atlas and CMS — and if we take one of them as accounting for the LEE the other is a genuine 3 sigma. The probability that this should go away is very very small and would need explaining if it does. Peter’s rumors do not undermine anything.
Matt,
I agree that both our choices about OPERA were reasonable, my point was just that there are other more important issues to consider than whether a result has been fully vetted by an experimental collaboration.
As for the “uncomfortable situation” I referred to, it’s a rather small amount of discomfort, mainly affecting the few scientists or PR people who may get questions from the press and public they have to say no comment to, as well as some annoyance for experimentalists who pay too much attention to blogs… On the other hand, there’s a very large number of people with an interest in this subject, and I see no reason they shouldn’t have access to this kind of news. As I think Mark recognizes, having some informed speculation distributed pre-announcement is likely to even generate more public interest in the actual announcement.
As for my polluting the “pristine data analysis”, I just don’t see any evidence that people are going to do their jobs any differently because of what has recently appeared on my blog. The pressures to get it right and done in time are the same no matter what information about what they are doing leaks to the public. There’s a legitimate argument that sometimes leaks from one experiment could affect the analysis by the other. Actually, late last November I was hearing about a signal around 125 GeV from one experiment, held off saying anything about it until I heard independently that the other experiment was seeing a signal near the same mass. In the case at hand, everyone knows exactly what channel(s) and narrow mass range to look at, as well as what last year’s data there looked like.
Well, it’s not for me to say if the amount of discomfort is small or not.
As for distributing information beforehand — this is a bit silly, it’s not as though there isn’t already plenty of interest in the Higgs. Your blog post from yesterday will do virtually nothing to increase interest in the Higgs that isn’t already there.
You may not see any evidence of problems, that is true. If such damage did occur, however, it would of course be kept secret… especially from you.
And this is where proving a case becomes difficult. I would suggest that you might consider being a bit more cautious.
a) Disclosing information that is incomplete, and not approved even internally within the collaborations, risks sharing information that should not yet be shared. Once such information is released, effectively unblinding data (internally within a collaboration or externally from an independent collaboration) before final checks are performed, there are certainly risks of influencing the answers obtained. This should not be done with other people’s data by bloggers who do not have full information about the internal analysis procedures they might be affecting.
b) on using one data set for significance and another for the look-elsewhere effect — this was for a “rule of thumb”, not a statistics theorem. And the only reason I treated them asymmetrically is that I know the details of the 2011 data now and I don’t know the details of the 2012 data yet. In principle you might be able to do it the other way (but that may depend on how the 2012 data is presented.) The right way to do it (eventually) will indeed be to combine all the data (under a particular theory hypothesis) and apply the look-elsewhere correction, and demand 3 sigma. Personally, given how often results turn out to be wrong, I would insist this happen from two independent experiments that confirm each other’s results, or that this happen in one experiment using two-photon and four-lepton results alone, since there is less room for error.
c) I left out the numbers because I have no interest in playing around with numbers that aren’t final and that don’t have errors properly accounted for; I have science to do. As for how likely this is to give us a “discovery” — do we really think that 4.5 sigma is not a discovery and 5.5 is a discovery? I’m not a believer in precision statistics (you can get different answers depending on what question you ask), and even less a believer in arbitrary decisions as to how to treat precision statistics. You can call these arbitrary decisions “scientific” if you want. In the end the only thing that matters is getting things right. When our significance reaches 10 sigma (which it will) then we don’t have to discuss these arbitrary decisions anymore. In the interim, these arguments as to whether there is or isn’t a Higgs discovery or just evidence, based on the exact numbers of sigmas, don’t actually accomplish very much.
This “subvert the scientific process” prissiness is just hilarious.
The only people who really care about the results already know, via their own professional rumour mills, pretty much what is going on.
The only other people in the world who care are people like me: amateur science enthusiasts, washed-out graduate students, and technically inclined bystanders of the type any academic professional who cared about the political realities of funding his/her field would go out of their way to engage with. even to the awful, awful, awful, unprecedented level of letting slip bits of excitement and scandalous, terrible, non-peer-reviwed rumours.
Faster than light neutrinos were really good for physics. Higgs rumours are good for your employment and for your grad student’s prospects and for the public funding of incomprehensibly complex endeavours that no lay member of the public will ever comprehend except at the human drama level. Anything that gets anybody outside the experiments to give a damn about the way the universe works — and how science really works, warts and all — is good for humanity. These are massive, expensive, huge, public-funded collaborations, and human members of the public are interested in more than what a consortium Press Office approves for staged revelation.
Pursed-lip tut-tutting about proper processes and the sanctity of Phys Rev B does nothing for anybody.
Save the secrecy and the dramatic revelations for blinded data analyses, where it actually does have any effect upon The March of Science.
Obviously I don’t agree with you. Inadvertently damaging blinded data analyses (and introducing other types of dangerous biases) are indeed one of the main things I worry about.
You can call it prissy if you like. Doing things right sometimes requires doing things that others view as prissy.
Matt,
In this response and in the one to Jolly Joker I question why you are abusing the term “blinded” when I don’t think you are so ignorant as to not know what the word refers to. You know very well that “blinded” data is data that has been “blinded” to the people doing the analysis themselves by the use of some transformation, and that this method is being used by the LHC experiments. The information on my blog is obviously about the unblinded results of the analysis, seen after the “opening of the box”.
If I were to be “effectively unblinding” someone’s blinded data analysis, yes that would be extremely unethical (as well as practically impossible and kind of insane…). There is something unethical going on here. It’s not my behavior, but your abuse of technical language in order to attack me.
You are right that the term “blinded” is very complex in this context; that is the reason I used a shorthand. It would take a long post to describe how these data are analyzed. I agree I should find a middle ground for explaining this.
However, I talk to people inside the experiments and they tell me your understanding of what you are putting on line is simply wrong.
My target is mainly the people who gave you the information. However, I would request, on behalf of the particle physics community, that you not take risks with irreplaceable information.
On what grounds do you know the rumor that you reported is safe for release?
The term “blinded” is not “very complex” in this context. It has a specific meaning as a specific tool used to get precisely the numbers we’re talking about. Instead of admitting what you did (and apologizing for it, which would be a good idea…), you change the subject and try and attack me on another front. This is pathetic.
Peter — as far as I understand, you’re not correct about the simplicity of blinding in the LHC Higgs analyses. It’s not so easy to do the simple blinding you can do in some experiments. I don’t have time for this discussion today.
You’re over-reacting, but if you want an apology for using the word “blinding” without a modifier, here it is: “I am sorry, Mr. Peter Woit, for suggesting that you unblinded an experiment.” I do not have any knowledge that you did so. And as I have said before, I have every confidence that you would not knowingly release information that should not be released.
My question remains (and you have not answered it): what gives you the knowledge that you have not released information that should not have been released? You are not a full member of both ATLAS and CMS; how did you obtain the information that everything that you released was in fact safe for release?
Actually Matt, it’s “Dr. Peter Woit”, a relevant distinction since, despite what you seem to believe, I do understand what I’m talking about. The arrogant belief that anyone who disagrees with you must be some sort of incompetent idiot who sadly just doesn’t understand is one of the least attractive aspects of lots of people in the particle theory community, and I’ve seen far too much of it in recent years. It’s an endemic pathology in the string theory community, with some string theory blogger poster boys. One thing about being in the math community is that you learn the importance of being very clear with yourself about what you know and what you don’t know. You are unlikely to ever believe this, but I do have some idea of what these experiments are doing. Not only that, but I also know what I don’t know, and how extensive that is.
The information put out on my blog was (intentionally) very limited, essentially just the news that both experiments are seeing a similar sized signal to last year’s in the gamma-gamma Higgs channel near 125 GeV, which conclusively indicates there is something there. The only thing that any LHC experimentalist who read my blog learned was (assuming that he or she hadn’t already learned this from a friend/colleague, you’re very naive if you think this is not going on) was that the competition was seeing (in the first part of the data) something similar to what they were seeing, which in turn was similar to last year’s data. This is very unsurprising information, and I don’t see how it would significantly affect how they do their job.
There are situations you could imagine where information leakage out of one experiment would actually be a problem, for instance if one experiment was seeing something surprising and unexpected. In that case I would be far more careful about revealing whatever information on this came my way.
Fine: “Dr. Peter Woit” If you prefer that, I don’t mind.
Well, as I said in past comments, there’s no way to determine who is correct here. I am sorry that you are not willing to be reasonable.
As for knowing what you are talking about — well, that’s obvious, Peter. You know a lot, but unfortunately, not quite enough.
By the way, I don’t see anything in my comments which suggests you are an “incompetent idiot.” If you were, I’d ignore you.