Talk:Self-organized criticality

From Wikipedia, the free encyclopedia

This article is within the scope of the following WikiProjects:


This article may be too technical for a general audience.
Please help improve this article by providing more context and better explanations of technical details to make it more accessible, without removing technical details.

Contents

[edit] Major rewrite

I've just made a major (read: total:-) rewrite of the page. Hope I haven't trodden on any toes; I just felt that not only did the page need expanding but that I could put some of the points in earlier versions better in my own words. Feel free to flame me on my talk page if offended. I do have quite a few SOC-related research interests, so it's possible some aspects of my rewrite may be too POV (though I've tried to avoid it). Advice on this appreciated.

Although broadly my aim was just to give an expanded picture, I have deliberately removed certain things said in the previous version. Here's a list of what and why:

  • The theory of self-organized criticality (SOC) claims that whenever a self-organizing dynamical system is open or dissipative, it exhibits critical (scale-invariant) behaviour similar to that displayed by static systems undergoing a second-order phase transition. — I don't think any of the serious proponents of SOC have suggested that "whenever" anything. But in any case this particular statement is not true. Example: take the BTW sandpile model and introduce dissipation into the local exchanges. It satisfies all the conditions listed in this "definition" but isn't critical.
  • Examples include avalanches e.g. in sandpiles — This is a confusion of the sandpile/avalanche metaphor to visualise SOC models, and the actual real case of avalanches (whether in sand or other materials). In fact the results from avalanche experiments are pretty ambiguous (see e.g. Turcotte's 1999 review).
  • This is related to the self-organization of cellular automata. — This is misleading. Many cellular automata display interesting emergent properties but most are not self-organizing. Example: Conway's Game of Life displays emergent complexity, but change the rules even slightly and it significantly changes the behaviour of the system. Self-organization implies robustness with respect to design variables.

I have expanded the reference list but deleted the link to Amazon since I don't think it's our job to tell people what bookshop to go to. However, I do plan to update the reference section so that all references to scientific papers have a link to the online copy of the article (original and preprint, where possible).

I plan to make a couple of further updates in the near future, most importantly, to give one or two examples of SOC algorithms (I'm thinking BTW sandpile and Bak-Sneppen evolution model). When I really have time (not now, not now) I'll try to add some handy graphics to help readers visualise what's going on and to show examples of the sorts of scale-invariance that result.

Comments eagerly awaited! Peace and love, — WebDrake 03:33, 7 October 2005 (UTC)

Great work, WebDrake! I've done a minor copy-edit, but I think your article is excellent.

Although it's correct as it applies to what was written, I don't entirely agree with your comment above on no one asserting "whenever": I think Per Bak's How Nature Works was criticised, at least initially, for his claims about the ubiquity of power laws (see [1] for a sample along these lines). But that's a small point. Thanks for your new version. -- JimR 06:41, 8 October 2005 (UTC)

Thanks very much for the nice compliments. Copy edits are always welcome. :-)
To the issue you raised, I think the criticism of PB was that it was perceived that he had claimed that whenever a natural system displays power laws, it is a sign of SOC, though I don't think he actually said or meant that; what he said was that the widespread occurence of scale invariant structures in nature pointed to a common dynamical mechanism (for which SOC was a viable candidate). There is an objection to this argument in the review you cite, which points out that many quantities follow a Bell curve (e.g. IQ and height) yet no one would claim these have a common dynamical origin. But actually, I think they do — it's just that these dynamics are so general that they can have vastly different physical embodiments.
Others interpreted the title of his book as being a claim to have explained everything (whereas I read it more as being a call to arms to start thinking about the question of "How Nature Works", with SOC being an exciting example of how to do so). Much of the criticism I've read of him seems to stem more from not reading or understanding what he said (probably often because they were so put off by his willingness to make blunt criticisms of others), and reading things into his words that aren't actually there.
To the direct issue of why I objected to the phrase in the earlier article, AFAIK neither Bak nor anyone else made any sort of "whenever" claims about what kinds of dynamics would display SOC. (Probably this is one of the reasons why Jensen's (1998) acronym of SDIDT [Slowly Driven Interaction Dominated Threshold] systems is not widely used, apart from unwieldiness. It gives you an idea of where SOC may emerge, but doesn't guarantee it.) Nobody has given any sort of "bottom-up" definition of SOC along the lines of "SOC occurs whenever the dynamics are like this" — but the phrase I objected to outlined SOC along those lines. — WebDrake 20:37, 9 October 2005 (UTC)

Further to the above, I've also slightly revised one of the paragraphs JimR edited, the 2nd para of the Overview, since I didn't quite feel comfortable with the way that was rephrased (sorry:-). Also, in the 3rd para, I was not comfortable with the removal of the word "some" from "some reservations" regarding SOC with local dissipation. The reservations are very minor and do not concern SOC as a whole, just some types of dynamics; SOC with local dissipation has not really been contentious in the scientific literature for the last 10+ years. So I put "minor" in brackets before the word reservations. WebDrake 22:01, 9 October 2005 (UTC)

I like "(minor) reservations". My edits were intended as stylistic not scientific: I only removed the "some" to avoid a second occurrence of that word in close proximity. Ditto for "possible" in the previous paragraph, but you've put that back :-)
With regard to criticisms of Bak, I think they may have arisen partly because of a lack of clarity in How Nature Works: as far as I remember, at no stage does he give an explicit definition of SOC — nor as you point out above did he specifically state a hypothesis that untuned scale invariance is always the consequence of a particular mechanism. On the other hand, his many examples of systems building to criticality and then exhibiting a power law did seem to lead people to think he was asserting a theorem of sorts — which they disagreed with. In other words, as you say, they perceived he was asserting a "whenever" statement. It would of course be nice if there were a theorem :-)
The first sentence of your article does give a clear "top down" definition of SOC; on the other hand it doesn't posit a specific "bottom up" mechanism. So it's not likely to meet the same criticisms. However, you do call it a "mechanism" in a couple of places further down, without going into details of what the mechanism is. Maybe that can be addressed. -- JimR 11:11, 10 October 2005 (UTC)
Hmm. I understand your issue over "SOC as mechanism". I'll think about it and how to phrase things better. "Self-organized criticality" and "self-organized critical dynamics" tend to be used interchangeably in the literature, which is the context in which I was writing the paragraphs you refer to. Probably the best thing is to do a slight rewrite of the opening paragraph. After all, in the sense of SOC as a description of theoretical models, it is pretty much proven for at least some of them that the critical point of the system is an attractor. So I'm probably wrong to give a definition which refers just to the observed phenomena. — WebDrake 23:31, 10 October 2005 (UTC)
OK, see what you make of the updated version of the intro. I've also added links to online copies of all journal articles referenced. :-) — WebDrake 00:58, 11 October 2005 (UTC)

I like it! The definition as "having a critical point attractor" is elegant and succinct. You then relate the attractor to the absence of tuning, and the scale invariance to the criticality, sensibly leaving the details about power laws and fractals till later. The new second paragraph then explicitly acknowledges the lack of a general theorem or guarantee (so far), thus drawing the teeth of the criticisms directed against (mis?)perceptions of Bak's implications. It's good. -- JimR 11:39, 11 October 2005 (UTC)

Great. Thanks for the little typographical/wiki link edits, that's been really helpful. I think we've got both intro and overview pretty sorted now. :-D
The next few things I want to do are to write up a couple of models as examples of SOC dynamics (I think BTW sandpile and Bak-Sneppen evolution model are good choices). I'd also like to have a hack at the self-organization and power law articles, but I'd like to chat a bit more with the existing writers on that since I think any major changes there would be more likely to cause a ruckus — self-organization, at least, is much more disputed and less well-defined than SOC. I guess you know the people who have worked on that stuff somewhat, so do you reckon you could bounce them into some discussion? (Get them to look over this article too.:-) Best, — WebDrake 16:01, 12 October 2005 (UTC)
Thanks for the feedback, WebDrake. You have good plans, but I'll respond on your talk page since we're going off the SOC topic. -- JimR 17:25, 14 October 2005 (UTC)


[edit] Examples of SOC dynamics

I've added a section on examples of SOC dynamics, with links to other pages rather than take up a load more room here. That can also allow for the individual model pages to get relatively lengthy. I've created the model pages as stubs so if anyone wants to write them, go for it! :-) — WebDrake 02:12, 15 October 2005 (UTC)

[edit] Further related pages

I've just created a stub page for the Oslo ricepile experiment, with a reference to the paper published in Nature.

In addition we need to create pages on the Gutenberg-Richter law and the Omori law for earthquakes.

Further ideas welcome! — WebDrake 23:36, 17 October 2005 (UTC)

[edit] Categorization

Now that Category:Self-organization exists and it is a subcategory of most of the categories Self-organized criticality is in, what is the rationale for keeping SOC in additional categories? Karol 07:58, 18 October 2005 (UTC)

You'll have to guide me on this since I'm a relative novice to the question of how and why categories should be used. My choice was related to two points: (a) is it appropriately classified? and (b) if someone is browsing through category lists, how easy is it to discover relevant and interesting material?
Essentially my rationale is that while classification is important, it is also important that material be highlighted as soon as it becomes relevant — else there's a danger of users having to descend through an endless tree of categories.
It's for the latter reason I decided to include the higher-level categories in SOC since I felt that someone who is interested in SOC will not necessarily be first and foremost interested in the feature of self-organization. There's also the question of the genealogy of SOC being a bit different since for example it's not too highly related to chemistry or cybernetics.
But I'll follow your guidance on this. :-) FWIW I don't think limiting SOC to Category:Fractals and Category:Self-organization would be too inappropriate. — WebDrake 13:50, 18 October 2005 (UTC)

UPDATE: After thinking about this further I've removed several of the higher categories. The remaining ones are Category:Applied and interdisciplinary physics, Category:Fractals and Category:Self-organization. I know that the first of these is a parent category of self-organization, but I think in this context it's important, since its use as an applied and interdisciplinary physics theory is one of the key aspects of SOC. Good compromise? — WebDrake 16:48, 18 October 2005 (UTC)

I'm not aware of the most common way categories are used, but I doubt that alot of people just browse them and look for things to read (I don't, unless I get really bored :D). I think it is more natural to enter a category from one of the pages it contains, with the aim of seeing what else it contains. Returning to the case at hand, the categories you had before and the ones that are there now seem quite fine - I say this because they are more or less consistent and because these categories aren't yet overloaded (I am very keen only on keeping Category:Physics slim for now). Also, as it is now, self-organization itself seems to be organized more poorly than SOC (so many categories!). One more note - SOC is probably much used in phase transition studies (I think so only, 'cuase it's not my field), but it's not related to Category:Phase changes in any way right now. Karol 20:55, 18 October 2005 (UTC)
Small retrospect. I fear you may have felt some pressure from my side as to the categorization of SOC. After removing it from Category:Physics, I only meant to notice that it might not be optimal. I write this, because I am grateful that you improved the article. After all, categorization is a second hand issue compared to the article itself. Karol 21:08, 18 October 2005 (UTC)
Naah, no pressure was felt. I listened to your reasoning and felt it had merit; your contributions have been useful both for the article and for me as a new Wikipedian. Glad you like what the article has become! :-) — WebDrake 00:16, 19 October 2005 (UTC)
P.S. "Phase changes" is probably not the best place for SOC. The critical phenomena associated with phase transitions are kind of the opposite of SOC — in a phase transition the critical point is only reached by highly accurate tuning of a control parameter, but in SOC it's an attractor and robust with respect to parameters. If there was a "Critical point phenomena" category, that would be a good one. WebDrake 00:24, 19 October 2005 (UTC)

[edit] Applied and interdisciplinary physics

To repeat the above (since some people keep trying to remove it), it is understood that Category:Applied and interdisciplinary physics is a parent category of Category:Self-organization. The choice is deliberate. SOC is clearly related to self-organization, but is not limited to that area of interest: it has wide applications in interdisciplinary studies. The categories are chosen to reflect both these aspects.

If you want to remove the category, please give a good reason on these talk pages first. —WebDrake 13:40, 22 November 2006 (UTC)

[edit] "The Neo effect"??

Where did the comment come from that SOC is sometimes referred to as "the Neo effect"? I've never heard of this term and was unable to find any reference to links with SOC except the comment in this article. So, I've removed it for now, but if anyone can convince me otherwise, I'll let it be put back in. —WebDrake 09:54, 19 July 2006 (UTC)

[edit] Smalley et al. (1985) and Katz (1986) papers

Can the person who made these edits please identify themselves and give some justification for this edit?

I am going to rewrite these edits on the following grounds:

  1. This article is about self-organized criticality, not the sandpile model. The concept of SOC was put forward by BTW and not before.
  2. The edits suggest bad faith on the part of BTW, which I see no reason to assume.
  3. I actually have the Smalley et al. and Katz papers, and it's not at all clear that the models are identical. Similar, yes.
  4. Turcotte wrote a review of self-organized criticality, referenced in the article, and he does not even mention the papers in question. Given that he was a co-author on the former, you'd think he might have done so if there was really a priority question.

I agree that we could mention, in the discussion of the history, that some earlier similar models to the sandpile model had been published. —WebDrake 12:09, 19 November 2006 (UTC)

Further to above, I have reverted the introduction but added some extra material elsewhere. —WebDrake 13:08, 19 November 2006 (UTC)

[edit] Bak-Sneppen references

My thanks to the anonymous contributor who added the Bak-Sneppen references. I'm going to move the Sneppen interface depinning reference to Bak-Sneppen model where it is more appropriate. The Bak-Sneppen reference itself I think should stay as it represents the introduction of the second general kind of SOC dynamics: extremal models as opposed to threshold/avalanche models. —WebDrake 18:16, 8 February 2007 (UTC)

[edit] Integrate and Fire Neuron Network with SOC properties - References?

Dear Webdrake, could you complete the model section with a literature reference on SOC in networks of IaF neurons. Many thanks.