DePriest, Greg (NBC Universal) wrote:
> I understand your frustration and hope you'll feel my pain.
> 
> There are a couple of things missing in your critique.  
> 
> One is the history of ALTO, its arbitrary exclusion of content
> protection, and the total inability of anyone to articulate the reasons
> for it having done so other than to say a decision was made (over the
> objections of content creators) and that's that.

Greg, reverting a decision made some time ago is not the problem -- it
happens so often in the IETF that we have RFCs defining procedures for
that. The problem is that there is strong consensus that there is no
effective way for asserting content legality in the protocol the ALTO WG
is supposed to specify.

Let me try and summarize the many discussions that drove the group to
such a conclusion. The protocol will be used between applications and
some kind of servers to exchange information related to the topology and
the policies of the underlying network, to let applications make
better-than-random decision when they can choose among a large set of
endpoints to connect to. The mechanisms for content protection
considered for this scenario -- at least those the group was aware of at
the time of the discussions -- are all based on some sort of content ID
sent by the client (i.e. the application) to the server; however, while
the server could use such IDs for optimization (e.g. to pass additional
endpoints he's aware of, most likely caches), because the content could
be virtually everything (other than pirated movies, also YouTube-like
user generated content, porn of any kind, or even, in the case of Pando,
large email attachments), it would not have any reasonable means for
asserting their authenticity. In other words, if a client sent a fake ID
in a request -- say, the ID of the King James version of the Bible if
not a random token -- the server would not be able to detect it without
relying on huge databases (that would be a nightmare to maintain) and
DPI solutions whose costs would immensely exceed the savings due to the
reduction in cross-domain traffic.

Now, since the group agreed to reflect such a conclusion in the charter,
this discussion is currently out of scope for the ALTO WG. However, if
people are aware of other content protection solutions not considered to
date that would integrate well with ALTO, I will be happy to discuss
them and to help investigate opportunities to extend the current charter
as per RFC 2418. But that has to be done off the ALTO mailing list.

> As for the role an ALTO server might play in the piracy ecosystem, I'd
> welcome that discussion.  Your second paragraph minimizes it as ALTO "is
> an optional service."  
> 
> Yet it seems to be a law of nature that options must, over time, become
> standard features. So I worry.
> 
> But my larger point is that it's just wrong to leave open the potential
> to expedite delivery of any pirated content.

Any improvement in the network has the potential to expedite delivery of
pirated movies, as well as communications among terrorists, exchanges of
pedo-pornographic content and many others. Unfortunately nobody can
address them all at the core of the technology; the approach engineers
are used to following and that proved to be quite successful at least in
the designing of the Internet consists in defining building blocks, as
simple and extensible as possible, upon which they and others can then
implement solutions for any sort of higher level problems. And I'd say
that this is the right approach for ALTO too, not because a decision was
arbitrarily made months ago, but because so far the WG has shown strong
consensus to do so.

-- 
Ciao,
Enrico

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

_______________________________________________
alto mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/alto

Reply via email to