Please help me explain P2P to Mark Cuban
Mark Cuban recently posted a series of commentary on why ISPs should block P2P traffic. Mark Cuban talking about P2P is like me talking about, um, sports team management, which might explain why a lot of people who read his commentary just laughed at him and moved on, but Mark followed up with a post that was highly indicative of the "I'm rich so I must be right" mentality where he asked if blog comments have value (Mark, I don't mean to offend you with these comments, I'm just telling it like it is).
Now, most people took this post as an opportunity to kiss Mark's ass with nice comments, but being the young punk that I am, I decided to call him out. Deep down between comments 81 and 100, I said:
Are you posting this because you made a bunch of stupid comments about P2P and everyone disagreed with you? You are saying now that when people disagree with you, they are flamers and their comments don't count? Why don't you just come to terms with the fact that you are wrong and go out there and educate yourself? — Comment 81
Surprisingly, I got a direct reply from Mark. I have to give him props for this; I might not agree with his stance here, but he's a good blogger for actually reading comments and replying directly. His reply:
actually, for this post, if they dont agree with me, they are wrong. Im still waiting for someone to post actual facts proving me wrong. Feel free to show me a paper or data saying P2P is efficient for the last mile. Sending and receiving the same bytes over and over from a peer that only asked to get it once… and then having other peers on the same segment there as a back up doing the same thing not too efficient. — MC
Well, Mark has a point. Saying someone is wrong, and actually explaining why, are two very different things. Mark deserves a serious explanation rather than any "you suck and you are wrong" comments.
So help me out here, everyone. Post an actual, factual, charts and numbers explanation here, or give me a link to one. Trackback even. My explanation follows, though I admit it doesn't have any numbers:
Say Google just released, um, Google Desktop Search version 3.0 and they are hosting it on their server in Mountain View, California. I am a Comcast customer living in Fairfax, VA and like everyone else, I want to download this program (I don't use GDS, but this is an example).
If I download it from Google's server, my ISP has to negotiate the traffic to send a request to California and download the file from California, which is on the other side of the country and involves a lot of talking to other servers/nodes on the web graph/tubes.
Alternatively, if I get on a P2P network and download it from other users who live near me, who are all Comcast subscribers (and one of them already did the work of getting the file from Mountain View), then my ISP only has to negotiate the traffic between me and the other clients in the network. All the traffic stays in Northern VA and there's less traffic for Comcast to negotiate. Also, I would probably get my file twice as fast since I would be downloading small packets in parallel from multiple clients rather than downloading the same number of packets in serial from Google's server, and by getting the file download finished faster, I then free up network resources for other users to download stuff, like pages on Mark Cuban's blog, for example. It's a win-win situation in 3 different ways.
I know this is a bit oversimplified, but I'm hoping others can validate my scenario here since this is how P2P is supposed to take the load off the client-server model.
Now one might argue, "well, people are downloading enormous files from each other and that's getting in the way of my blog/news/christianmontoya.com reading," and in some ways, that might be true. This, however, is not a problem created by P2P. Users who use a lot of network traffic are probably paying extra for high-speed broadband plans, so the ISP has already done the market work of putting users into tiers to charge them for the level of service they want. If the ISP oversold their capacity and some users start getting less-than-expected speeds, that's the ISP's fault.
Now one might argue, "well, the answer here is to put people on limited plans. You only get to upload X gigs per month, after that you start getting slammed with high fees." That might sound like an attractive idea, but the market already went through that phase. There used to be all sorts of pricing models; Internet by the hour, by the Gb, etc. Phones right now still have data plans with limits. But as ISPs struggled to kill each other through friendly competition, they all moved towards flat-rate unlimited plans across the board. That's where the market ended up on its own and imposing limits to change that is only going to lead to the market eventually leveling itself out again. Maybe one ISP will impose Mark-Cubanesque limits on P2P traffic, and then a competing ISP will offer P2P-friendly traffic, and you'll have one ISP for the Cubans (ha, that's a pun) and another ISP for the P2Pers and eventually I suspect that the P2P ISP will be more successful and there goes Comcast's pseudo-monopoly. This is entirely hypothetical but I really think it would happen.
I hope that's a good explanation. If I'm wrong about anything, tell me why. If you can offer more information, by all means, please. Thanks in advance.
I don't use P2P - either for free or illegal downloads - but to legislate against it because some cock (telling it how I see it) wants faster broadband? Ridiculous.
Why not ban all traffic from people playing online games? I don't play WoW so it's not my problem. I don't want to see people's holiday pictures - let's ban the people uploading shitty snaps of their spouses in saucy beachwear. While we're at it, let's cap those who like to surf LOLcats because all them pictures must use up precious bandwidth!
Stupid.
If I can't play my online games I don't know what I'll do for fun
The main issue is that client-server and p2p are two different solutions to the problems, and you can't attempt to compare the two without narrowing down what you are looking for, in terms of application. There is PLENTY of technical specification and personal experiences out there on why p2p in file distribution is more effective: even more so if you narrow it down to bit torrent.
It's the reason why I can download the new Radiohead faster from any of the public or private trackers vs direct download from Radiohead. Server-client interaction is gets worse as more and more clients attempt to download from the server. P2P, especially BitTorrent, gets better the more people try to download.
MC is just being lazy. There is significant evidence that points to why p2p is better at file distribution than the traditional client-server model.