Comments on: Bump Hunting (Part 3??)

By: Philosophia Naturalis #8 « {metadatta}

Philosophia Naturalis #8 « {metadatta} — Sat, 31 Mar 2007 15:08:00 +0000

[…] Having huge meetings and partying like rock stars isn’t everything, though. Among other things, the physics community (just like any other) has its share of scandals, politics, marketplace tactics, things of that sort. Sabine Hossenfelder, for example, has recently blogged about the problems of treating the scientific community as a marketplace, while Julianne Dalcanton’s post on physics’ “cult of genius” definitely touched a nerve among readers. Meanwhile, Clifford Johnson has shared his views on recent events regarding an imprisoned theoretical physics grad student. (And of course, there’s the media aspect of things: John Conway recently picked up on his two previous posts on the search for the Higgs boson to blog about the unexpected media response.) […]

By: Gordon Watts

Gordon Watts — Fri, 16 Mar 2007 07:28:21 +0000

John — glad to see this post! You are actually getting quite a bit of flack over here after the NS article. I’ll point people to this.

By: tyler

tyler — Tue, 13 Mar 2007 20:19:52 +0000

“that same group” == e.g. you people

so thanks

By: tyler

tyler — Tue, 13 Mar 2007 20:18:56 +0000

As an extremely interested layperson I hope the tendency towards increased scientific discussion in (and between) blogs continues and even increases. Tommaso’s point about this discussion “filling the gap” between formal papers and the magazines is a good one. Physics papers, at least ones of value, are largely over my head; I can glean only meager understanding, and probably much misunderstanding, from reading them, though I find the effort interesting and sometimes enjoyable. Popular science written by journalists is filtered through the minds of writers and editors who, for the most part, seem to understand things much less well than even I do.

Just as I only read “pop sci” books written by actual working scientists, having been burned one too many times, so also am I only really interested in reading online or in magazines from that same group. My own lack of rigorous understanding is more than enough filter already, without adding someone else’s ignorance into the signal chain upstream of me.

This blog is a perfect example of the sweet spot - some posts are technically over my head (I am in the same boat as George re: your bump-hunt graph) but only partly so. I feel like I always at least grasp the essence of what’s under discussion, and the increasingly interactive nature of the discussion is quite exciting. I am more interested in the science posts than the cultural ones, though I am more qualified to comment meaningfully on the latter…;o)

As for the open discussion being misread or spun: it’s either accept and try to mitigate the reality of that happening, or give up the discussion. Don’t do that. This, I think, is one of few areas where the blog format actually brings something new to the table. It’s a step forward.

The future of scientific discussion looks a lot more like Cosmic Variance than it does New Scientist, or at least I hope so.

By: Brian

Brian — Sun, 11 Mar 2007 22:37:48 +0000

Arun - Thanks for alerting me to the article in the Times: I really enjoyed it.
Paul Valletta - By a “bump” they don’t mean an aberration in a particular datum. As you point out, the more data, the more individual oddballs one is likely to have. By a “bump” they mean that an unexpectedly high PERCENTAGE of the data has a certain characteristic. The more data one has, the more likely it is that the percentage of the data in the sample with, say, energies between 159.0 and 161.0 GEV is close to the percentage of data with that charactreristic in the underlying population (reality) rather than being a mere statistical fluke. It makes sense that the more data one has, the more accurate one’s assessment can be. I hope that helps.

By: Hal Porter

Hal Porter — Sun, 11 Mar 2007 22:19:53 +0000

John:

As a sometime science writer, let me commend you on the simplicity and clarity of your post. This is a real contribution.

Also, like others, I can appreciate the excitement and frustration that such tantalizing results offer. While most of these signals may be spurious or artifacts of some sort or just statistical anomalies, some probably really are a first manifestation of a deeper reality.

It must be a very exciting time for all of you, especially since it seems to me that you are all a generation younger than I am.

By: Arun

Arun — Sun, 11 Mar 2007 20:05:57 +0000

Paul - re -

“A small amount of data looked at, would theoretically have less chance of a fluctuation, than a vast pool of available data?”

To give a simple example, suppose I was trying to find out whether a coin had heads on both sides, and the only way to do so was to toss the coin and observe one of the sides. Let’s assume that the coin is fair - namely it lands on either side with equal probability.

If I tossed the coin 3 times and saw heads all three times, it is possible that the coin landed on the same side all three times (one time out of eight trials of 3 tosses).

If I tossed the coin 21 times, and saw heads all three times, the probability that the coin landed on the same side all 21 times is approximately 1 in 10 million, and I’d be much more confident that both sides of the coin bears heads.

By: George

George — Sun, 11 Mar 2007 16:39:16 +0000

I’m just a layperson with an interest in the beauty of the thought processes are at the core of science’s attempt to understand the complex structure of the world we live in. I started reading CV because it puts me closer to the process of discovery by eliminating the filter of interpretation used in the popular press.

I’m interested in the Higgs boson because I think it’s a beautiful idea even though I have no clue about the how or why it might work or even exist. So, I started following along the discussion here, the first post left me looking at the graph and asking myself, huh, what bump? As the topic developed over the following posts I think I came to at least crude understanding of what the graphs were displaying and what the fuss was all about.

I’m assuming that the detector allows one to sift through the collision debris by detecting where it registers on the instrument as a function of its mass or energy. I’m also guessing that the decay process is producing signals from events that are either not part of what is being looked for, or that might otherwise be random, unexplainable or unknown. This causes noise in the results where the target particle, the ‘bump’, must be discerned against what is in effect a fuzzy mean. So if the amount of data analyzed is increased, the fuzzy background should tighten up within the 1-2 sigma bands and if there is a target particle present, its signal should become more distinct, a bump with a higher sigma.

Of course I could be totally wrong, but it is interesting thinking about it.

By: Count Iblis

Count Iblis — Sun, 11 Mar 2007 14:30:58 +0000

Does the number of x-sigma anomalies in High energy physics experiments follow the Normal distribution? Is there a publication bias (null results being under reported) causing a false impression that there are more anomalies than statistically expected, which in turn could be generating more speculative theory papers?

By: Paul Valletta

Paul Valletta — Sun, 11 Mar 2007 14:00:02 +0000

Quote:”The team had found a signal which, in particle physics had a 2-sigma significance- a 1 in 50 chance of being a random fluctuation. Normally,to merit new particle status a signal must be significant to 5-sigma-where there’s only a 1 in 10 million chance of it being a fluctuation”.

I would have thought these odds are incompatable?

A small amount of data looked at, would theoretically have less chance of a fluctuation, than a vast pool of available data?

More data should have more “bumps”, and all should adhere to the correct predicted signal.

The more you look for something, the more chance you have to find it, it’s when you are ‘not’ looking for something and it exists at a specific location, (no matter how much available data one has!)..that this is the experimentalist dream?

Surely, finding something twice, in a small amount of data, is far more exciting than finding soething once in a vast pool of data ?