Jump to content

Douze points de levier

From Meta, a Wikimedia project coordination wiki
(English) This is an essay. It expresses the opinions and ideas of some Wikimedians but may not have wide support. This is not policy on Meta, but it may be a policy or guideline on other Wikimedia projects. Feel free to update this page as needed, or use the discussion page to propose major changes.
Translate

Les douze points de levier pour faire evoluer un systeme de en:Donella Meadows' (twelve places to intervene in a system) appliques a Wikipedia:

  • 12. les nombres (constantes, standards, parametres) - la longueur des articles, le nombre de liens, le nombre d'editeurs
  • 11. La taille des buffers et autres stocks stabilisateurs relativement aux flux
  • 10. La structure des stocks de materiel, des flux et des noeuds d'intersection - confort d'utilisation, capacite du serveur a assumer les requetes, temps de connexion
  • 9. Les temps de delai relative au temps de changement du systeme
  • 8. Les boucles de retroactions negatives, relatives aux exces qu'elles essaient de corriger - vandalisme, etiquette, guerre d'edition
  • 7. Les boucles de retroactions positives - ethique, communite, gouvernance, [[:w:fr:Pens%E9e_de_groupe|pensee de groupe]]
  • 6. La structure des flux d'information (qui a ou n'a pas acces a quel type d'information) - Main Page style, person DTD, spacetime DTD, Special:Recentchanges, pages de discussion
  • 5. Les regles du systeme (contraintes, punitions, blocage) - blocage d'IPs, [[:w:fr:Wikipedia-la_neutralit%E9_de_point_de_vue|Neutralite de point de vue]], les listes de suivi, contributions, 'suivi des liens'
  • 4. Le pouvoir d'ajout, de modification, d'evolution, d'auto-organisation de la structure du systeme, e.g. les sysops, les escadrons de la mort, les trolls dans la legion des trolls, dans l'[[:w:fr:Wikip%E9dia:Historique_du_Wikipedia_fran%E7ais|histoire de Wikipedia]], et dans les [[:w:fr:Auto-r%E9f%E9rence|auto-reference]]s ([[:w:fr:Wikip%E9dia:Auto-r%E9f%E9rence|Tout ce qu'est Wikipedia]])
  • 3. the goals of the system, e.g. avoiding becoming What Wikipedia is not
  • 2. the mindset out of which the goals, rules, feedback structure arise, e.g. the "punishment works" mindset of en:G. W. Bush, that of NPOV itself
  • 1. the power to transcend paradigms, overthrow an elite, change a structure.

Where does writing articles fit into this paradigm?

giving content to articles is one of the goals, that is 3.
Writing articles has impacts all the way up and down the scale, so does reverting them, editing them, deleting them. Writing an article changes the number of articles in the system and number of links in the system to other articles (the "citation count" of "what links here" for instance) - which change the probabilities of flows from one area of the wikipedia to another. Then, the more interesting material there is in one area, the more people interested on this topic or value systems will come and edit articles in turn

And where doest awarding sysop status or taking sysop status away fit ?



There are neither nine nor twelve points, I'm a little confused here. :)

eh, good perception !
w:Donella Meadows first proposed a 9 level system in a meeting, then further refined it in a 12 points system.

There is a positive feedback loop at work:

  1. Google indexes Wikipedia
  2. This makes Wikipedia better known to readers
  3. This brings new editors (and potentially knowledge)
  4. This creates more and better articles
  5. This creates more links to Wikipedia, increasing our PageRank
  6. Then we get yet more google hits, so more editors again, etc

However, the increased database load from having more readers and editors increases connection times. This may drive our usual contributors away. In other words, points 9&10 act as a constraint on the system.

Another problem with getting many new editors is that it may bring more basic vandalism. This would take up all the time of sysops trying to cope with vandalism and new comers welcoming and teaching. This behavior happens when a system starts changing much, much faster than its negative loops can react to it.

-


Where does writing articles fit into this paradigm?

giving content to articles is one of the goals, that is 3.
Writing articles has impacts all the way up and down the scale, so does reverting them, editing them, deleting them. Writing an article changes the number of articles in the system and number of links in the system to other articles (the "citation count" of "what links here" for instance) - which change the probabilities of flows from one area of the wikipedia to another. Then, the more interesting material there is in one area, the more people interested on this topic or value systems will come and edit articles in turn

And where doest awarding sysop status or taking sysop status away fit ?



There are neither nine nor twelve points, I'm a little confused here. :)

eh, good perception !
w:Donella Meadows first proposed a 9 level system in a meeting, then further refined it in a 12 points system.

Wikipedia system is a growing system, where information accumulate, to reach (an impossible ?) goal of being the repository to all human knowledge.

Positive and negative here don't mean one has positive or negative final effect, it refers to the direction of the change.

The negative feedback loop tends to slow down a process, while the positive feedback loop tends to accelerate it. That's all.

Negative effect will promote backward direction, and positive effect will promote forward direction. Of course, forward and backward are all relative. It depends of the goal. It depends whether the goal is to progress, or regress, or tend to stability.

In a system going forward, the negative loop will tend to promote stability (stagnation in this case). The positive feedback loop will tend to speed up the growth (ie, the accumulation of knowledge).

But, both can, without control, get so much influence that they may completely break up the system. Be they meant to speed it up, or to slow down the process. Both can get wild.

Example of positive feedback loop going wild

a nuclear reaction, such as in Tchernobyl
eutrophication of a lake, which first lead to increase of productivity (biomasse increases), to lead to an oxygen-deprived system where microorganism cannot assume biodegradation of dead matter any more, to end up with a dead lake where anoxic conditions caused the death of all oxygen-dependent life

Example of negative feedback loop going wild

deforestation of Amazonian forest. While many people would imagine soils in these forests are rich, they are usually not (except in places where Indians have been making sort of tertres of accumulated wastes). Remove the trees for agricultural production. The bare thin soil will not resist long with few plants on it to tropical rain falls, and will suffer erosion. It will quickly loose the little organic matter it has, resulting in ever less species thriving on it, resulting in even naker soil...to the point where the mother rock is reached and no vegetation grow any more.

When Google is indexing Wikipedia, it is making Wikipedia better known to readers, and can bring new editors, so potentially more knowledge. That's positive ! Then, more editors mean more articles, more links, so more google hits, so more editors again. That's a positive feedback loop !

However, in indexing, it is strongly increasing connexion times, hence driving usual contributors away : that is a positive feedback loop going wild !

And here, the bottle necks are 10 and 9

If Google brings too many new editors at the same time, it may brings more basic vandalism, and may take up all the time of sysops trying to cope with vandalism and new comers welcoming and teaching. To the point they are not authoring articles any more. That is a positive feedback loop going wild ! (though...it may make sense to prevent some sysops to edit...)


essentially right. As long as you view the definition of feedback from an electronician or mathematician perspective. Which is clearly not a biologist perspective. Fact is I think you view it from the input and output perspective only. Not from a system perspective. Biologists see feedback both from the output perspective and system perspective. It might be that our definitions are different.

"negative feedback" is feedback that acts so as to reduce its own cause. An example is body temperature - if I get too hot I sweat more, and this negative feedback reduces my temperature.

Notice the wikipedia definition is saying that a Feedback is a process whereby some proportion of the output signal of a system is fed back to the input, in order to change the dynamic behaviour of the system.

This is quite right on the most basic system, but not taking into account that sometimes, this is not the output that is fed back. The output signal can be fed first to another system, which output will act on the first system. Or it can be transformed by a function.

It also says Feedback may be negative, thus tending to reduce output, or positive, thus increasing output.

That is forgetting most systems work with threashold values. And until the theashold is reached, the system is changing, not the output. Besides, syneresis also impact system, without impacting output necessarily so.


Positive feedback is feedback that acts so as to increase its cause. There is no such thing as a "negative feedback loop going wild" - what you describe is a positive feedback loop with negative consequences. Similarly a nuclear meltdown is a positive feedback loop with negative consequences. MyRedDice

Negative feedback is for neutralization/regulation of the system (homeostasis), while positive feedback is an amplification. Both feedbacks are not linear, nor parallele to each other. Which is why a tiny little thing like a butterfly can have a huge impact, for its little effects may be amplified by a positive retroaction. I agree there may not be negative feedback loop going wild, but I still maintain a positive feed back loop can certainly get wild by an un-controlled amplification. Which is why in most biological systems, processus are controlled both by positive and negative feedback loop (well...usually by several). Often, the process is slightly too slow that required, so a positive feedback loop is accelerating it more or less for it to have the necessary effect. But, there is usually a negative feedback loop acting as a guardian, to ensure no excess amplification occur. Typically, heart rate with vagual system.

I may not use the right words to express myself, so I will probably not convince you, but it will take much more for you than just saying "you are wrong and wikipedia is right" to convince me :-)

- feedback : stabilisation

+ feedback : amplification


Extract

Positive feedback loops are sources of growth, explosion, erosion, and col lapse in systems. A system with an unchecked positive loop ultimately will destroy itself.

And

The most interesting behavior that rapidly turning positive loop can trigger is chaos. This wild, umpredictable, unreplicable, and yet bounded behavior happens when a system starts changing much, much faster than its negative loops can react to it.