Talk:Strategy for Information Markets/Network Externalities/Networks and Network Structure

Knfritze review
I liked how this page was broken down into the different types of networks and how we use and analyze them. I thought the basic definition of networks was well defined but I do believe it could be shortened a little because most people know what networks are. I really enjoyed that each section gave an example of the specific network that section was discussing at the end. I am still a little unclear about the abstract network section. I might have missed it but I did not see an actual definition of an abstract network. I think we could work on this section being more concise and better defined, but the examples at the end were helpful.

Knfritze (discuss • contribs) 23:49, 6 February 2012 (UTC)


 * We'll be relying a lot in this course on the abstract network idea, so making that clear is important. I have tended to dodge around other issues of network structure, but they can be very important so finding other sources to fill out that material would be an excellent addition to the text. TDang (discuss • contribs) 23:17, 16 February 2012 (UTC)

Pwningviolation review
It did have good formatting and sectioning. Abstract network structure takes a bit of reading to really understand what it's saying. I'm still not sure where associations within a larger network encompasses it, but it could be edited down more if that definition is satisfactory.
 * Pwningviolation (discuss • contribs)


 * The abstract network idea is one we'll be using very heavily, so working to make it more clear would be good. TDang (discuss • contribs) 23:15, 16 February 2012 (UTC)

research on why people join Facebook
Check out this article: "How Facebook "Contagion" Spreads". That kind of thing is about network structure--that it's not just the number of people which is important for many networks, but what kinds of relationships those people have. TDang (discuss • contribs) 22:18, 2 April 2012 (UTC)

TDang review April 2012
I'm reviewing this version. I'll likely be more critical than complimentary, because (a) that's the way I am and (b) that's what will help improve things. Please don't take the criticism-over-compliments to mean I have a wholly negative view.

Make sure to check the all-purpose review thoughts as well.

TDang (discuss • contribs) 19:57, 1 May 2012 (UTC)
 * Introduction is currently one sentence. It would be great to have more to it than that, whether by adding new text or moving stuff to the introduction from other sections.
 * "complex systems"--this has a particular meaning in math (and applied math). I'm not sure if you're using it in that way, so it's probably better to phrase differently.
 * "When looked upon in a general point of view, it can be seen that most networks are relatively similar in nature."--We will often make an abstraction that networks are similar to one another, but I don't think I would want to say that abstraction is always safe. Different kinds of networks can behave very differently.
 * Virtual and Real Networks--You can leave this (if you do so, please reference it) because this is how some people talk about networks. I don't think it's a very useful distinction for analysis. If two people are talking on the phone, does it matter if the network is defined by actual wires connecting versus software saying which information packets should be read where? I think that the virtual-versus-real idea is either outdated or trying to get at something else perhaps one-sided direct network extrnalities versus multi-sided indirect network externalities? Anyway, as I say, you can leave it but if you can figure out a way to improve it, that would be terrific.
 * Network Structure Variations
 * sponsored versus unsponsored is a relevant issue for networks, but isn't really a matter of network structure. Is there a better place this should go?
 * Likewise with open versus closed (although part of what could happen with an open network versus a closed network is that the owner of a closed network would have more power to dictate the network structure while an open network, the structure might change as people join, leave, or extend the network).
 * " An example of an open network is the public radio because anyone can enter this market."--This is probably not a good example for two reasons. First, it's not obvious that public radio should be thought of as a "network" in the strong sense we're frequently using for this course. Second, public radio isn't open in that I can't just go start up a public radio station whenever I want, I need permission from the FCC.
 * "An example of a closed network ... is a public park"--I would probably agree with this, if I had to take a stand, but it's fuzzy enough that I don't think it makes a good example. Public parks (versus, for instance, private shopping malls) typically mean people have civil rights to be in the public space, and that they can only be ejected for certain, limited reasons. A mall can eject someone just because the mall feels like it (unless that "feeling like it" is based on certain kinds of illegal discrimination).
 * "Only organizations with a strong hold upon the market can exercise a strategy of control over their network." This, and the remaining two sentences of the paragraph are attempting to pack a whole lot of information in, and not really getting it. I'm not sure it's about network structure. Whatever it is is probably more like 2 paragraphs with clear explanation than 3 sentences.
 * Abstract Network Structure
 * This section is tricky, and not something we spent much time on in the course, but would ideally be the core of the page. The question about an abstract network is not "should the network be abstract"--the network is what it is. The question is can we model this network as having abstracted network externalities and still have the model be useful?
 * In the course, we almost entirely used "abstract" models where we just looked at the size of the network (or the expected size), and the value of the network to an individual just depended on that size. That is an abstraction, it leaves out:
 * There are specific people who I want to be connected with, others I don't.
 * Even among those I want to connect with, some connections are "stronger" than others.
 * Some people are more highly-connected and/or more influential than others.
 * Some networks let you connect more directly, while some you have to go through an intermediary.
 * ...probably lots of other things.
 * We leave those out because it lets us talk about models for lots of different things as if they were essentially similar, and we hope that they are. That's the "abstract", the decision about how to model. If some of those elements really change how the network behaves (maybe there's 2% of people who are highly-connected influencers and so things respond much differently to their joining the network than some less-connected person), then the abstraction will lead to a bad (not useful) model.
 * I should be getting more of the flavor of this in this section and I'm not (although the references do look like they're getting at those ideas).
 * If there are graphs you can find which help to clarify these ideas, that would be a huge benefit to this page.
 * Two-Sided Platforms--One interesting thing about multi-sided platforms (for which I don't have a reference handy, sorry, and it might be hard to track one down so you might have to ignore this...) is that abstraction is more likely to be a good modeling choice for them. This is because it's more likely to be true that the different sides of a two-sided platform care more about the size of the other side and less about who specifically is on the other side. And, even more so, when you take it to the second step and think about how people on one side think about others on the same side, they only care about how many are on their side because that's what influences growth on the other side.
 * Analyzing Networks--These look like potentially very useful references, but it's not clear that you really understand what they are saying in selecting what quotes to give and what to do with them.
 * Overall--The writing here is mostly good, but too wordy.
 * Some sentences like "In order to grasp the multiple facets of the structure of networks it is important to understand what a network actually is." don't add any real content--they are setting something up, but it's better to jump right to what's being set up.
 * A bit more analytical examples of different network structures would be great--again, with pictures if possible.
 * One possible resource is http://www.cs.cornell.edu/home/kleinber/networks-book/. This will be too much to absorb in a short about of time, and any of the more-sophisticated pictures from the book you should avoid using in case it's improper, but you might be able to get some worthwhile ideas. I'd suggest glancing at chapters 11 and 20, and chapters 2,3,4 for background if you feel you need background.
 * References
 * Sonamine looks like a "solutions provider" kind of service, so stuff from them needs to be viewed with suspicion. However, at the bottom of that web page is a reference to an academic paper which hopefully is more complete, trustworthy and contains similarly useful information?
 * The Hagiu reference looks great.
 * Try to reference the Hagiu paper and whatever replaces Sonamine with "inline references" similar to what's used in the Analyzing Networks section.
 * For all the inline references, give more complete references so it's not just a link, but title, author, etc.
 * The earlier material on the page needs references.
 * "In a study done by the Haas School of Business..."--need full reference to that study.