October 9th, 2005 by Hafiz Noor Shams
Project Petaling Street is all good and cool but it suffers one complication – how many participants are too many participants? In short, what is its carrying capacity? Will it fail if the limit, if any, is not observed?
Allow me to explain further.
PPS, or to be precise, its ping portal, PPS Pings – hereafter PPS – is a first-in, first-out list (FIFO). The list itself has limited slots.
If there are five slots in a clean list and if person A is the first pinger, he will sit on top of that list. Later, when four other persons ping the list after person A, A will sit at the bottom of the list while the four persons will sit on the top four slots. It follows that if a sixth person pings the list, person A will get the boot. In short, if there are n slots, the person on top of the list will be out of the picture when pinger (n+1) pings.
All this assumes strictly one-to-one relationship; a person leads to a ping and a ping leads to a person. This assumption is not true (enthusiastic posters, as a PPS founder calls them, like Kahsoon and Otis, is a proof) but it definitely simplifies our model. Once we’ve laid out the model, then you and I could relax the restriction considerably.
With FIFO explained, let’s talk about rate. Let’s also assume that each slot receives x visitors per time unit. Here, I don’t think one-to-one assumption is critical but for consistency’s sake, let’s just assume one-to-one relationship.
Assume further that each ping is pinged into the list at a rate of a ping per time unit. Holding time unit constant, the rate is dependent on the number of pings. Subsequently, a ping goes down the list at that rate. Therefore, person on top of the list will be out by the (n+1)th time unit. Also, amount of visitors of that ping will be x(n).
So, imagine that a slot receives 60 visitors per hour and there are five slots. With this setup, a ping will attract 300 visitors throughout its lifetime in the list. Now, what if a ping switches a slot per minute?
That would make each slot a visitor and a total of five visitors throughout the ping’s lifetime. What if, the rate is a ping per second?
Oh, boy. A ping gets a sixtieth of a person; total a twelfth of a person. Ugly.
Ceteris paribus, the rate of which a ping goes down the list depends on the number of pinger. However, the more pingers there are in a timeframe, the greater the rate and as such, less visitors for every ping. The dilemma is, PPS wouldn’t be so successful if it weren’t for all those pingers. I would probably explain how more pingers leads to more visitors later. But I do have a feeling that at first, more pingers leads to more visitors until at one more point, any additional pinger will lead to less visitors.
This brings us back to the question how many participants are too many participants? In short, what is its carrying capacity? And when the carrying capacity is known, should PPS administrator put a cap on number of pinger? Or at least control the rate?
These are normative questions – there is no right or wrong. Nevertheless, by answering this, you might realize where you sit in economic freedom spectrum with controlled economy on the left-hand side and free market on the right-hand side. Of course, basing your political belief on this is absurd but hell, its fun.
So far, PPS Ping rules state that:
PPS will not tolerate ping flooding. Multiple pings within 7 minutes of each other is considered ping flooding, no exceptions, even if you are an “enthusiastic blogger”. It does not matter whether the content is duplicate or different from the previous ping. It is your responsibility to ensure your blog does not multi-ping PPS in consecutive fashion.
This alone is an attempt to control the rate. Does this mean the administrator distrusts free market?
I don’t know but it seems that rule is not strictly enforced.
However, I’m content to say that this is an attempt to prevent abuse and maybe, even market failure. The current DoS attack is similar to market failure there where is too many pings that visitors for each slot goes down to zero, albeit PPS fails due to bandwidth flight (capital flight? LOL!) first before market failure actually takes place.