Napster Controversy

This morning's Slashdot has a rather sensationalized headline about Gnap, the GPL'd client for the Napster protocol. Here's some more background information on the controversy.

Ok, there seems to be a lot of heat about the napster thing. I was there during a bunch of the discussions, so hopefully I can shed some light on what really happened.

Ryan Dahl (yes, he really is a distant cousin of Roald) started hacking Gnap, a GPL'd client for the Napster protocol about a month ago. From my discussions with him, he picked the project because Napster was cool, and because it woul be a fun Gtk+ project to hack on. His motivation has never been to create controversy or seek publicity, and now that it's all over slashdot, he feels a bit caught in the middle.

A few days ago, he got a sourceforge account, and put an alpha release of the code up on the page. A few other people saw the code, and interest started growing.

Apparently, Shawn, the main coder for Napster, heard about Gnap and sent a personal letter to Ryan requesting that he not release the source publicly until Napster had a chance to make their server more robust against flooding and similar denial-of-service attacks. The letter also referenced a usage agreement, implying that it forbade the development and/or use of a reverse engineered client.

Ryan sought the advice of the Gnome community, and decided on writing what I think was a really nice letter, declining to remove the source and pointing out that the real problem was that the server had weaknesses, not the existence of a free client.

Shortly after this, Shawn and Ryan had an irc discussion in which Shawn reiterated his belief that the existence of an open source client would invite abuses of the Napster server. He also pointed out that Napster plans to roll out some anti-abuse features in a future server. Ryan asked for a timeline, but Shawn basically said he couldn't say when that would be. This was actually a fairly civil discussion, actually.

Well, about 0100 UTC Thursday (ie, wednesday afternoon pacific time), a good old-fashioned all-out IRC war broke out on #napster on EFNet. Anyone who so much as mentioned Gnap was kickbanned from the channel by somewhat overzealous moderators. It shouldn't be hard to predict how people reacted to that :)

So, at this time it was pretty clear that the situation was pretty volatile, and before long a number of Gnome people (myself included) and Shawn and Jordan Ritter (nocarrier), and Slashdot's own Justin found themselves on an irc channel discussing what to do. At first, tempers were pretty heated, but things actually calmed down before long. Ironically, one of the things Shawn was most upset about was his perceived threat that the Gnomers would post this whole thing to Slashdot in an attempt to smear his company. We tossed a draft back and forth (quite a lot less sensationalistic than what actually got posted to /.), but ultimately got stuck on the fact that Shawn was unable to speak for Napster, Inc., regarding its position on Gnap. We encouraged them to get an official statement out quickly, but ultimately left it there. In any case, they made it clear that they respected that Ryan had legal rights to continue Gnap development, and that they were not threatening him.

Then, of course, 0919 UTC Thursday, Roblimo had to post a most inflammatory headline on Slashdot. I'm happy to see that most of the responses were reasonable and moderate. In any case, I can imagine that a lot of people are confused, which is why I am writing this.

So, what's the aftermath? Both Napster and Gnap get a lot of publicity. There is definitely no stuffing the genie back in the bottle, so the development of Gnap will probably go forward full steam ahead. Slashdot solidifies its reputation for irresponsible journalism. The world goes on.

And what of the issue of open source clients for proprietary protocols? Basically, if there are security issues of any kind (even denial-of-service), the responsibility really rests on the designer of the protocol and the implementer of the server. There is nothing that you can do to stop abusive clients. Trying to keep the protocol secret is pretty much doomed to be ineffective. My guess is that Gnap's release will contribute a fairly small amount overall to the abuse that the Napster servers get (and apparently they have already been getting some well before the Gnap release).

I suspect that the broader issue is who gets to control the protocol. In today's world, companies just don't get to lock other people out from implementing their protocols. The Internet is not a small, controlled, closed world. It's never been explicitly said, but it's not hard to guess that the real reason for Napster's concern over Gnap is the extent to which it might detract from Napster's control over advertisement delivery. As more and more new advertiser-driven Net protocols are introduced, it's likely this won't be the last wrangle over a free software implementation.