Beware of the Train

Last night a friend of mine retweeted this:

.@marzillk What energy!? Some completely unobservable thing? Energy is simply the ability to do work. Nothing more, nothing less.
-- rhysmorgan

It turned out that this was in response to the following:

@AlabasterC My sister recommends homeopathic aconite and tapping various energy points. It is helping a bit.

When questioned further marzillk usefully clarified:

@rhysmorgan Spleen meridian boosts immune system. It's under ribs on the left hand side - find the sore point, then tap while slow breathing

@rhysmorgan I know it sounds mental but I can genuinely feel it doing something.

Now, there's so much wrong with that that I don't know where to start. But I do know where not to start: by criticising her terminology. I fired off:

@rhysmorgan right now, @marzillk's making observations and you're bitching about terminology. Pop quiz: which is more scientific?

I want to expand on that a bit.

When I went off to university ten years ago, one of the questions I wanted to answer was "what is this thing called energy?" I still don't have a good answer to that, unfortunately; if asked now I'd mumble something about Noether's Theorem and Hamiltonians of closed systems, so I am at least confused on a higher level. But let's accept rhysmorgan's definition "the ability to do work" for the moment (and pretend not to notice that we haven't defined "ability" or "work"). There's still a problem with criticising marzillk's use of the word "energy": the physicist's definition isn't the only possible definition of the term. The OED gives seven definitions, of which only the final two are related to the thing that physicists talk about. Furthermore, the word "energy" has only been in use in the physicist's sense since 1807 (and the following year, someone proposed using it for what we now call "momentum"!) So if the practitioners of traditional Chinese medicine (or anyone else) want to re-use the term "energy" (or, for that matter, "meridian") to refer to something else, I have no problem with that in principle.

What I do have a problem with is the extraordinary ignorance of biology betrayed by marzillk's use of this stuff (and if she thinks that it's irrelevant that she works in educational publishing... well, I'm afraid I disagree). I've encountered a lot of skilled martial artists who claim to find qi (or something very similar) an essential component of the way they think about their art, so I'm not willing to totally write it off (perhaps it could be made to work as a high-level abstraction over some low-level biomechanical details), but really, tapping a point over your spleen to cure a cold? How on Earth does that fit with the (extremely well-attested) germ theory of disease?

But you know what, I could be wrong. Maybe tapping your spleen somehow makes it produce more monocytes or something (would that even help with a viral infection? IANAB). But that's the great thing about science: we don't have to rely on arguments from theory. Instead, we just do the experiment; if the experiment contradicts the theory, then the experiment wins. In this respect, marzillk was being more scientific than rhysmorgan, with her "I don't understand it, but I've tried it and it seems to work", approach. XKCD nailed this one: '"Ideas are tested by experiment." That is the core of science. Everything else is bookkeeping.'

To be sure, it's a pretty dodgy experiment. Only one subject, no control, no blinding, results measured by subjective feelings of wellness. One could imagine a much better experiment, in which a large number of subjects were randomly assigned to two groups: one group would receive meridian-tapping therapy from a trained master of the art, and the other would have random points on their body tapped by actors who could spin a convincing line of bullshit about the mystic significance of what they were doing. Then you measure how long it takes everyone to get better from their colds, and see if there's any statistically significant difference in the recovery times of the two groups.

I had a brief look, but couldn't find anyone who'd done that experiment. A fascinating and well-known 2007 study did what I described for acupuncture and lower back pain, though, and found that the actors had a greater success rate than the acupuncturists!

Contrariwise, this study evaluated "Meridian three-combined therapy" (thread embedding, bloodletting, and tapping/pressing) for the treatment of psoriasis, and found a small but significant improvement in effectiveness compared to conventional treatments. However, I'm assigning that limited weight in this context, because (a) it didn't just measure tapping, (b) it was for a different condition, (c) it was painfully (hoho) obvious to the patients which group they were in, and (d) it was performed in a Chinese cultural context, and we know (from experiments!) that cultural context is very important for the placebo effect.

So, anyone with greater scholarship skills able to find anything more relevant?

Update: rhysmorgan responds "I guess my beef with her referring to it as energy was that her form of energy doesn't actually exist." I take his point, but "existence" is kinda problematic in this context. Physicists' energy is an abstraction, a consequence of the time-invariance of physical laws. You can't measure it directly, nor capture it in a pure form. In what sense does it exist? But the theory of physical energy makes useful predictions: we can do calculations with energy and arrive at correct, numerical predictions of what will happen when we perform experiments. This is the important thing. To take a more abstract example: when you learn about the conventional underpinnings of calculus, you discover that there's no object in the system called "dx" or "dy", and that statements involving them are shorthand for more complex statements about the behaviour of limits. But the theory as a whole allows you to manipulate these non-existent objects and arrive at correct results. Analogously, the important question about meridian-tapping is not "does it refer only to directly observable things?" but "does it give correct predictions?"

For her Master's research, wormwood_pearl has to analyse some video. Not that much video, in the scheme of things - a couple of hours in total - but she has to decide and note down what's happening in every one-second chunk of said video.

Ladies and gentlemen, I give you the non-stop, fast-moving, white-heat world of SCIENCE!

[The lab has licenses for some expensive-looking software that's meant to do this automagically and without human intervention, but needless to say it doesn't actually work.¹]

So she asked me if it was possible to script VLC to play a video for a second, then pause for a couple of seconds (so she could update her spreadsheet), play for another second, and so on until the video finished. That way she'd (a) get accurate seconds, and (b) save lots of time that would have been spent clicking back and forth. I said "Er, yeah, probably".

It turns out that recent versions of VLC do, in fact, have a nice-looking scripting infrastructure based on Lua, but I don't have, and can't install, VLC on my work machine. So (at addict_yin's suggestion) I tried it with mplayer instead, and after a bit of fiddling got it to work. The program's very simple, but it's possible that someone else has this need, so I thought I'd make it public.

If you too need to play a video in pause-interspersed chunks (of any length you like!), you can download Sleepy here. Installation and use is very simple, and described (at least for *nix systems - let me know if you want to use it on Windows) in the README. It requires perl and mplayer, both of which are free and run on damn near anything. If you think you can improve it, please fork the GitHub repository (because all the cool kids are using GitHub these days, right?), and then send me a pull request. All comments and criticisms are welcome.

Yes, this is all massive overkill for what's actually a very simple script :-)

¹ It could be worse. Apparently an important early experiment on foam formation required a grad student to make up a soap foam between two glass plates, put the assemblage on a photocopier, and then press "Copy" every few seconds for up to ten hours straight. As my friend Micah put it when telling this story, if a nodding bird could do your research better than you could, you've probably taken a wrong turn somewhere.

The Green Party of England and Wales have totally re-vamped their science policy, basically fixing all the issues that journalists were raising in advance of last year's elections. So if that was stopping you from supporting them, you can now re-evaluate that position.

Note that the Scottish Green Party didn't suffer from this problem in the first place.

Ten probes from the Venera series successfully landed on Venus and transmitted data from the surface, including the two, Vega program and Venera-Halley probes. In addition, thirteen Venera probes successfully transmitted data from the atmosphere of Venus.

Among the other results, probes of the series became the first man-made devices to enter the atmosphere of another planet (Venera 4 on October 18, 1967), to make a soft landing on another planet (Venera 7 on December 15, 1970), to return images from the planetary surface (Venera 9 on June 8, 1975), and to perform high-resolution radar mapping studies of Venus (Venera 15 on June 2, 1983). So, the entire series could be considered as highly successful.

Totally. Fucking. Awesome.

There was a British space programme, briefly. I've heard it said that it could have got to the Moon for a tenth of what NASA spent, with their throw-money-at-problems approach, but that it couldn't have done it for 1/200th of what NASA spent, which was its actual budget. Sic transit gloria Britannorum.

There is also an Indian space programme, of which I expect great things in coming decades.

Inspired by mrpjantarctica's comment on my last post, here's a first cut at a "which scientific topics do you understand?" meme. Currently, it needs a lot of work: it's heavily biased towards topics that I know about, and topic selection's a bit dodgy even within those. So, can any of you suggest some better topics, or things I've missed? I'd particularly appreciate suggestions from chemists and statisticians, because currently those sections just serve to show my ignorance of those subjects! Hopefully I'll be able to post a completed list later today, and then we can start spreading it around.

( Instructions and list )

Those of you who do scientific computation may be interested in this link:

Bye Matlab, hello Python, thanks Sage.

Short version: Sage is a bundle of all of the various numeric/scientific/graphing tools for Python, made easy to download and install (even if you don't have root access, apparently). According to the blogger linked above, it's now good enough to serve as a replacement for Matlab (and it has ambitions to be a replacement for Maple and Mathematica too, though I don't know how far along it is). It integrates with R, Gap, etc. And because it's Python, you get a real, high-level, well-designed programming language with a proper environment to do your coding in.

In other news, wormwood_pearl finished her Finals yesterday. As you can imagine, we're both pretty relieved. There's no tradition of meeting people outside Finals here like there is in Oxford, but I went along with a bottle of champagne anyway - and only then remembered that public drinking is against Glasgow bylaws. Drat it. We went out to lunch, the bottle went into the Department fridge, and we subsequently went round to her sisters' and drank it there while watching Layer Cake. Because we're that exciting.

There's a new Fine Structure story out, called Too Much Information. In addition, the Crash stories have been incorporated into their own subdirectory, called 1970-. Some analysis of TMI and its implications has already started in the comments threads of my previous FS posts, and I'll have to revise my own thoughts in light of it, but for now:

( Discussion, including spoilers )

By the way, if anyone else is planning on blogging about FS, please leave a link here, so we can all follow what each other are saying :-)

I've been reading some science fiction written by a guy called Sam Hughes¹, and posted on his website Things of Interest. Most recently, I've been reading his Ed stories, about Sam and his (fictional) genius flatmate Ed, who builds battlemechs in their basement and saves the world from annihilation on a regular basis, and to whom Sam plays a sort of bemused Watson. It starts out as a fun bit of wish-fulfilment à la early Sluggy Freelance, and turns into something rather more. Well worth a read.

But I don't want to talk about that: I want to talk about his novel-in-progress, Fine Structure, and where I think it's going. It's a collection of loosely-linked short stories whose connections only become apparent later on, somewhat like Trainspotting: I don't want to give away too much of the plot, but let's just say it has elements of Contact, Superman and Strata, in a refreshingly hard-sf style. And it's garnered positive reviews from no less a person than David Brin (search for "Power of Two"). It's probably best to read them in order, but I started with Power of Two and it didn't do me too much harm (and it's one of my favourites).

Edit: further posts on this topic can be found here.

( MAJOR spoilers. Go and read the stories first! )

A word of warning: I have literally lost days to browsing Sam's website. Interesting days, mind :-)

¹ Some of you might even know him: he was a maths student at Corpus, Cambridge a few years ago. Or you might have encountered him on some online community or other: he generally goes by the username "sam512".

It seems the link between depression and serotonin levels is far less proven than generally assumed. More here.

[BTW, if you don't already read bad_science, you should.]