Comments on: Interesting semi-logarithmic graph – YouTube Traffic Rank

By: zac

zac — Tue, 30 Jan 2007 03:26:23 +0000

Ah yes, Moti – you have nailed Web 2.0 on the head.

The social networking force has many implications for institutions of learning. The old education model may have a limited lifespan…

By: Moti

Moti — Mon, 29 Jan 2007 16:30:03 +0000

One of the important questions is: why?
In this case, why did YouTube enjoy this tremendous growth?

The reason is the “force” that lies beyond much of social networking (MySpace) , part of Google’s rise, and Windows/office domination. It’s network effects. The more people are in such a network, the more the network is desirable. Further more, if one is comparing two such networks, their size is an important consideration.

Thus, we can think of the network as accelerating and not growing at a constant speed. A good exercise for students is to graph various such networks and compare the graphs and deduce the strength of the network effect.

By: MathNerd

MathNerd — Sun, 28 Jan 2007 13:31:54 +0000

Yeh, thanks Zac. It’s a really interesting graph. I remember the semi-log graph from school days, but we never had examples like this.

By: zac

zac — Sat, 27 Jan 2007 09:45:53 +0000

Thanks Alan for your suggestion, which I have implemented at Log-log and Semi-log Graphs. BTW - I like your domain name! Very clever.

By: Alan Cooper

Alan Cooper — Sat, 27 Jan 2007 09:43:41 +0000

I love this example of the use of a semilog graph to emphasize the area of greatest interest in a ranking situation and will be using it in my precalculus class a week or so from now. I can also imagine that a loglog graph might be useful to get a view of high frequency small amplitude oscillations near the origin and/or big slow variations far away, but would be hard put to come up with such a topical example. Of course a loglog graph is also very useful for identifying a power law (which it converts to a straight line) and I would like to suggest that showing an example of this (perhaps with a non-integer exponent) might be a worthwhile addition to your page – especially since the loglog transformation of an exponential (which you demonstrate) is again an exponential so the only effect in that case is an overall scaling.
cheers,
Alan