Slashdot Mirror
Power Laws, Weblogs, and Inequality
scubacuda writes "Clay Shirky has written an excellent article entitled "Power Laws, Weblogs, and Inequality." Simply put, diversity plus freedom of choice creates inequality: "A new social system starts, and seems delightfully free of the elitism and cliquishness of the existing systems. Then, as the new system grows, problems of scale set in. Not everyone can participate in every conversation. Not everyone gets to be heard. Some core group seems more connected than the rest of us, and so on." A must read for anyone interested in the statistics, fairness, and power relations of blogging."
For something with a real distribution (either bell curve, or in this case power), a REALLY small number of datapoints shows you the pattern, if the datapoints were drawn randomly.
I would guess if you picked just a few 100, the graph would look the same.
In this case, what is more suspect is that we have inclusion of the biggest, most popular weblogs, which implies they were included by hand. Therefore the sample is probably not random.
In addition, there is no definitive list or way to calculate all blogs, so by defenition his sample comes from a subset of all blogs that were in some way listed or linked to. There are probably a whole strata of blogs that were guaranteed not to be in the analysis, because they werent included in whatever source material he drew his sample from.
Power law distributions can be seen with very small sample sizes. Sounds like you need to learn some statistics. Note: most college courses do not cover this subject well. Mathamaticians are only in the last 50 years developing the tools needed to model the positive feedback systems that often yeild power distributions.
You don't need a tethered PC to Blog. All one need is a PDA. There is a LiveJournal client for both the PalmOS and PocketPC OS for example, and probably for the Zaraus as well. And since many of these PDA's have wireless capabilities you can literally blog from anywhere in the world you can get a cell phone signal.
Now I agree that you would have to be an ultimate loser to want to blog right after getting laid, or out in public after some mundane activity but hey, to each his own.
Ours is not to wonder why, ours is but to Blog and die.
Mac OS X and Windows XP working side by side to fight back the night.
However! When you are trying to make inferences about the population like this, you must have a frame.
What does this mean? The question is: how can you produce a random representative sample when you cannot identify the population from which to take the sample? To obtain a sample you have to first produce a list of entities from which you must choose
Recently I was contemplating conducting a demographic study of such a hidden popluation, in that case enlgish language instructors in Warsaw, Poland. How was I to find the instructors? There was no listing, and even if there were some kind of directory, there is little reason to think that it would be representative.
so in the case of the blogs, if you can't locate certain blogs to include in your frame, then you can't include them in your sample, and so your sample is going to be skewed.
however, in any case, the skewing would just chop off those lesser known blog on the tail end anyway, reducing the extremity of the distribution.
one way of identifying such hidden populations is through the technique of snowballing, which basically is to take advantage of the connectivity of a network. you ask a blogger to name other bloggers, and you keep a list of those you identify. the problem is that a snowball sample is not random and thus not good for inferential stats.
Logic, macros, and more