Slashdot Mirror
P2P Developers Stand Up To Intel
Simone submits this article about two different visions of peer-to-peer computing - one from Tim O'Reilly, and one from the Intel corporation. (O'Reilly expounds further in his column). Watch and cheer from the sidelines as the mega-corps jockey to control the buzzword standards process, turn it into a useless mush, and are surpassed by protocols that work.
The W3C's effort was a lost cause the day Andreesen ran off to the left coast with the Mosaic code, closed it, and starting saying 'mine' with regard to the WWW.
And it didn't help when Microsoft started doing the same thing awhile later.
Tom gave a rather interesting talk at a BioInformatics Open Source Conference (same mob that's into www.bio{perl|java|xml|python}.org and generic tools for hacking the genome) a month ago where he did discuss some of the relevance of peer-peer. The essence of peer-peer is an basically lack of centralised control (something that quite rightly annoys corporations) and dynamic reconnectivity (create new services by adapting old). Since I was there, I've scribbled down a transcript of his talk which may be of some interest (caveat ... it's released under OpenContent but Tom should be given right of first proof to make sure I didn't take down his words in vain :-) so treat it as rough working notes until then). Basically we had the old point-point connectivity (think 1-1 e.g. ftp) of the old days, then the client-server paradigm (think 1-n e.g. http) currently. Now we have an arbitrary n-n connection pattern where the programming style is not as clear. Different services have different patterns of usage and new protocols/frameworks are currently being explored like BXXP. However, the value proposition is not gated communities (aka portals) but how many other groups find your services valuable (ie commons). You can't churn users through a limited set of data portals (cough*hotmail*cough) and influence/restrict their movement. Remember the basics of commerce is built upon the premise of an economic good which is excludable and rivalable and peer-peer sorta tweaks that model quite seriously (hard to stop another peer replicating your "stuff"). This becomes a little more interesting when you're trying to search a couple of hundred terabytes of gene annotations, ESTs, microarray data, etc. as you want to combine both completeness (to maximise success) and minimal covering set (to save costs).
Why are the big names interested? As ever, they want new drivers of growth (notice the PC market is becoming saturated). As for the buzzword du jour crowd, well that's what a cluebat is for :-).
LL
hehe, good point...
----
---- I made the Kessel Run in under 11 parsecs.
I think its great. It gives the suits some justification to throw money at technology, including geeks like you & I. Of course we all know its total BS (i.e. one of the main benefits of P2P is that it is virtually free.. how does anyone expect to make money off that??). But as long as the buzzwords attract venture capital and big IPOs, who cares?
----
---- I made the Kessel Run in under 11 parsecs.
The reason Intel, Cisco et. al. are interested is because P2P adds value to their hardware. A large portion of Intel just exists to create neat software that require more powerful server and client hardware. So Intel may indirectly make money off P2P, just like they made money off B2C, B2B, ASP, etc - by providing the underlying infrastructure.
----
---- I made the Kessel Run in under 11 parsecs.
Why does something have to be a money machine to be considered a success?
Napster is a company which has been given millions of dollars in venture capital in the hopes that it will reap profits for its investors. If Napster fails to give the investors a return on their investment then it has failed as a business, whether lots of people use it is immaterial, just ask the investors in CDNow, DrKoop.com, Pets.com, etc.
Second Law of Blissful Ignorance
I find it interesting...the article I had from Business 2.0 open on my desk when reading the linked article by O'Reilly runs much along this vein. Alas, since it is published in the 10/24 print issue and not yet published on the Web, I can't link to it.
It's more interesting to me to watch the consortia-standards group fight than it is to worry about any single fight over any standard. If there's anything that watchers of IT should know, it's that if technology standards stink, we'll ignore them. You would think that businesses would eventually kowtow to standards groups, as consortia-derived standards tend to have a shorter shelf life--driving costs up as a new standard must be developed. Independently-developed standards tend to have greater market acceptance and therefore a longer shelf-life--eventually driving down marginal cost and increasing profits.
Why the companies that are out creating consortia can't realize this is beyond me...
--
-- Geof F. Morris
Of course they should structure their comittee like the W3C.
After all, look how good a job the W3C comittee does at convincing MS and Netscape of the value of standards!
Standards work well for technical concepts where it's really in the best interest of everyone to do it the same way, and everyone realizes it.
But for new concepts where companies already have their own buzzword-compliant concepts to make them money at the expense of the general industry, standards efforts are doomed to fail!
If the corporations don't all get behind a standard the standard languishes [ie: firewire and memory stick] or dies [ie: Betamax]. Pay to play is the way big corporations think. It's the wrong way to do this and I am glad to see a few intrepid and fearless souls like Tim stand up. But... if we don't get a protocol 3com, Intel, and The Evil Empire [Microsoft] all support wholesale, it could die on the vine.
Don't get me wrong: a protocol just needs support, and not necessarily corporate support. Corporate support would just make the protocol available in a more widespread, fait-acomplis sort of fashion. It's ultimately users that decide protocols.
Pay no attention to the man behind the curtain with all your metadata.
I don't think they do. Say it with me, now -- "PEER TO PEER". It describes a network of coequals. The people to whom this notion appeals are not going to like a Oligarchy governing structure. "We love the idea of a peer-to-peer network of equals, and that's why we have to rule you like a king to build it!"
-----------
-----------
If you ever drop your keys into a river of molten lava, forget 'em, because man, they're gone. -- Jack
So, we have here a definition that makes servers talking to PC's a P2P relationship. . . hmm. . . don't ftp, and http, and nfs fall into this category? Of course they do. The truth of the matter is that the suits-and-ties-and-power-lunches-and-buzzword crowd are just now discovering the nature of the internet. . . That any computer can talk to any other computer (provided it has an assigned IP). So what all this P2P "buzz" is really about is just what high-level software protocol standards we'll use to facilitate peering different machines in a given situation. E.g. how will "my" e-commerce server talk to your credit-card validation service server to know that the customer can or can't purchase from me.
I do agree with you though that this is a lot of people spouting hot-air right now.
I know you understand what P2P is; perhaps you don't fully realize how it is different from the old-school internet model is.
The fundamental difference is obsoletes the "mainframe style" of computing which Sun thrives (i.e. of selling a few high powered servers) and instead has a lower-powered server at everyone's home.
See why Intel loves this?
Intel's fundamental business plan is two-fold: to get more people to use computers, and to get more people to use more powerful computers.
Intel is active in bridging the digital divide, making computers easier to use, and more accessible, attempting to achieve the first goal. But they are also active in trying to get people to use more powerful computers, and they are constantly developing new software technologies in order to demonstrate the need for more powerful computers. P2P is a glorious example of this.
I think Intel is imagining a world where every internet technology is peer-to-peer, such as e-mail, file sharing, and web technologies. It is fundamentally different from the current model because it gets rid of the powerful centralized server, and requires everyone to have a server.
Many people don't realize this, but internally, Intel considers Sun to be much more of a competitor than AMD. P2P is the one technology which has potential to make Sun irrelevant, because it makes big, centralized servers irrelevant, and replaces them with Intel desktop machines.
Michael
...another comment from Michael Tandy.
"Goodness me, how unlike the FBI to abuse the trust of the American public." -- The Onion
Or so old and bloody obvious that no-one before Intel thought that they could get away with making up a fancy name and marketing it.
Both Tim O'Reily and Intel are just blowing bullshit here. There is no such thing as "P2P", at least not in this new definition of it (that is the definition that they aren't sure what it is yet), and everybody except the press, the bullshitters, and the venture capitalists know it.
I mean, what exactly are Napster and Gnutella that makes them something new and revolutionary?
They are simply search engines (and rather bad ones compared to say Google) that can handle a high volatility of the hosts where the data is located. And the only reasons why this high volatility is present at all is that:
a) ISPs are doing their best to keep people from having static address where they could put normal servers. This partially because of the lack of IPv4 addresses, but mostly in order to maximise profits by having users that need servers require more expensive accounts.
b) The data they are being used to distribute is illegal and so non-volatile sites carying it are hunted down.
So, a) is caused by corporate stupitidy at the ISPs (many of whoom, I'm sure, will say they are P2P supporters now that it will help the stock price), and b) is because of the legal stupidity also known as copyright law (which has it's biggests supporters in companies like Sun, IBM, and Intel, all "P2P working group founders" according to the ZDnet story). There is no revolution here, just a clumsy workaround for a completely unnecessary problem. What these programs offer is simply the chance to download stuff that has been hunted off the web by lawyers working for the same corporate structure that now thinks "P2P" will be the next big money extorting machine for them...
I am one of the three or four lead designers/coders of one of the networks mentioned in the article, and I honestly couldn't give a rats ass how this working group is formed, or who controls it - at least until they can actually define what the fuck they are in control of, and what it is that makes this something new different.
(The project that I am a part of, Freenet, is trying to do something new, but what we are trying to achieve is only of the very surface connected to "P2P" as the money crowd sees it. And it will take years before we are ready (the Internet was given 25 years before the money crowd realized it was a prime rape target - why can't we have a tenth of that?))
Is anybody else tired of all these Three Letter Acronym buzzward fads that for the past year or two come long every few months and are proclaimed to be the greatest invention since blah and will become a market worth billions of dollars only for a few companies to IPO successfully just to tank a few months later.
Frankly I can't see how P2P is any less a meaningless buzzward filled, destined to fool a lot of investors, make a bunch of young CEOs rich, unproven, hype filled technology. Every time I open an issue of Fortune or Forbes I see some fool going on about how P2P is the next big thing and how Napster being so successful confirms this. Now unless I have been living in an alternate universe for the past year, Napster is not successful. They've spent millions of dollars obtaining lots of eyeballs, but the one thing the dot-comm massacre has shown us is that eyeballs do not necessarily transform into revenue. This reminds me of all the dotcomm evangelists who used to claim that B2C was the way to go because Amazon was so successful (even though they are yet to turn a profit and spend $150 million a year just to service their debts).
Second Law of Blissful Ignorance