|
Web3D Security Discussion |
 |
|
|
Join The Discussion
|
"web3d modeling (e-commerce)..."
JOSE_PARRA
|
|
|
|
|
|
|
|
|
|
|
Dateline: January 24, 2001
Every few months on the "big" VRML discussion list (www-vrml@web3d.org) someone brings up
the question, "Can I create content that is secure?" and the resulting answers brings
forth a religious stream of verbiage. I'd like this article to be a fair presentation of the issues
and views of people on both sides of the arguments. I'll present slightly edited versions of
posts (edited only for clarity and presentation) and conclude with some suggestions that might lead us toward
solutions that might make some progress.
I'd also like to thanks all of the participants for allowing republication of their posts. In addition
it should be noted that all of the participants are speaking only for themselves not for
any of their employers.
The latest round was sparked by:
At 05:18 PM 1/24/01 +0100, Jan-Olof Janson wrote:
Is it possible to protect VRML files from being edited but still being
able to being viewed in the VRML browsers (In the gzip compression
maybe?)? I there any "obfuscated" method that generate VRML code that
can't be structured by any automatic optimiser?
// Best Regards Jan-Olof Janson, Dynagraph
To which Cindy Ballreich replied:
Jan-Olof,
Not only is security not currently possible, It's unlikely to be
available
anytime in the future. Many people feel that mild security is not useful
and unless complete, unbreakable, perfect security can be achieved that
we
shouldn't even try. Others believe that security is unethical and all
VRML
files should be accessible to anyone who wishes to view or edit the
source.
This discussion always turns to methods of supporting copyright law (such
as watermarking). The primary issue is that an open standard (especially
one with open-source implementations) means that a determined cracker
with
the right skills has access to source code and will always be able to get
to the data. The text based nature of the basic VRML/X3D file adds to
this
problem.
My opinion (stated many times here) is that the lack of even mild
security
has hurt VRML in the past and will continue to do so in the future. I
believe it's a factor in the success of some of the proprietary web3d
technologies. I'm sorry this hasn't received more attention, but perhaps
there's no point.
Cindy
CKS responds:
There's no point in trying to come up with a purely
technological solution, but I wouldn't argue with the
importance of discussing the issue.
For instance: what exactly are customers looking to
protect. Content, right. And they don't want people
to pirate it. But that's not specific enough:
- Does the content have a lifetime? The Superbowl
video-feed is worth a lot more during the game than
after the game, for example. Copying after-the-fact
isn't so much of an issue.
- If the content contains built-in ads (like TV),
then the more piracy, the merrier. Any viewer is a
good viewer.
- If, like DirectTV, you resell content for money,
then you end up with very, very expensive custom
hardware. You've got to do this, even if you know
it won't always work, because it's your only revenue
stream. And you've got a neverending and very
expensive battle with the crackers to content with.
- If, like movie studios, you depend on reselling
content for long periods of time to great numbers
of people, then you end up with things like special
edition packages of Gone With The Wind. Everybody
already has it on tape, but not everybody has the
book + special interviews. (We actually bought one)
- Music videos are so hip they have a lifetime of
only a couple of weeks. There's not much time for
widescale pirating and distribution.
- Large scale commercial pirating operations are
the target of massive international law-enforcement
efforts.
- Etc, etc, etc.
In other words, it's not that copyright protection
isn't important, it's that there's no silver-bullet
technology that will solve the problems. The moral
objections come up when copyright holders push for
draconian laws that infringe on the rights of
everyone in a misguided attempt to exert an impossible
level of control. That harms everybody in the long
run, including the content producers...
-cks
Andy Best chimed in with:
I believe it is very important for the large scale commercial development
of
virtual worlds. If there is no solution on the horizon this will be one
factor forcing us to look at other technologies.
Some type of author key connected to a particular viewer might be a
possibility, although of course by necessity destroying the ability to
use
different plugins to view the content.
- andy
---------
andy best
co-founder &
chief creative officer
MEET Factory OY
Clayton C. says:
ok so what i see here is:
a) a security paradigm is needed
b) it must be global enough to encapsulate any size of project
c) it wont make vrml a binary format
{oh where has the vrml-cbf workgroup gone? }
d) must be able to be on any platform and viewer {conformance}
e) it must be granular enough to protect certain objects while other
ones can go right on through
so why not {psuedo code}
SecureTransForm {
acl_passwordfile "level5"
requires valid-user
encrypt_type BlowFish
children[]
transform
rotation
...
}
it seems to me any browser could use this sort of way of providing the
encrypt mech so objects could be encrypted
only ones in the securetransfomr would need to be processed like this
of course the main file would contain the encrypt type not good
wouldnt need it if everyone used the same one
which would be better overall for conformance anywho
which encrypt modules work
on all plats & in multiple languages
Oliver Mellet from openworlds adds: We (http://www.openworlds.com) have had similar requests from our
customers
regarding the integrity of the content that they create, and they
generally
fall into two broad categories:
1. How can we protect our content from manipulation by third parties?
2. How can we ensure that our content will not be pirated?
The first issue is easier to address, using some sort of open checksum
system or digital signature system. The second point is considerably
more
complex, because, as someone pointed out earlier, at some point the data
has
to be rendered, and is therefore susceptible to piracy. However, any
system
that seeks to completely eliminate the possibility of piracy is doomed to
failure (CSS, SDMI, MacroVision, etc...) for precisely the same reason.
So,
our goal would be to make it as difficult as possible to intercept and
reverse engineer content, eliminating all but the most determined
pirates.
Having said this, we also don't want to speak for our customers, so we'd
like to hear from content creators/providers about what they would find
useful.
Thanks,
Oliver Mellet
oliver@openworlds.com
Cindy says:
This response is typical of the many discussions we've had on this
subject.
A group of potential users with real projects comes to this community
with
a list of issues and requirements for meeting them. The community's
response is that the issues aren't real.
This is the wrong answer. There is a profound lack of understanding in
this
community of the issues surrounding the commercial use of 3D technology
on
the web. Projects are being lost to other technologies or lost altogether
because of these issues. Denying the they exist will not change that.
If this community doesn't wish to address these issues (because it
doesn't
like them, or it doesn't believe them, or whatever reason) it has every
right to ignore them. But don't insult the professionals that are coming
here asking these questions by telling them that their problems aren't
real.
The Facts:
- Your concerns and the concerns of your clients are real.
- There is no prefect security. If someone (like Christopher) wants to
apply enough effort, any security you use will be broken. (But you
already
knew that, didn't you.) Confidential information should *never* be placed
on the web - in any format.
- Open Standard/Open Source solutions (like VRML/X3D) will probably not
be able to provide the degree of security that you or your clients are
asking for. There are too many opportunities for people to take advantage of
easily available information to break the security.
- Proprietary solutions *may* be able to provide you with enough
security
to meet your needs. Ask lot's of questions, and question the answers.
Cindy
Len says:
I guess it is not that the issues don't exist,
but that the answers always depend on the requirements
as stated. The facts you present are facts. The
means depend on the level of protection required
(none all the way to abstinence).
Again, a variety of means is applied to open formats:
- No protection
- Zipping.
- No zipping and copyrights in place.,
- Zipping and copyrights in place.
- Encryption
- Binary
- Digital signatures (similar to restricted access)
- Restricted access and rights to access and redistribution clearly stated.
Any others?
There are also combinations of the above. Complete
security is abstinence. After that, only in a closed
room and shoot your partner afterwards. On and on,
but it comes down to the extent to which one must
go to protect that which they want to protect. The
means are there. The only agreed on means at this
time for VRML is zipping.
Umm... these are pretty much the same answers we've
seen in past discussions. Are there new requirements
or are customers unhappy with the answers? The binary
seems to me to be the only thing VRML doesn't enable.
Are you asking for a binary?
Len
clbullar@ingr.com
Next page > Sandy says... > Page 1, 2
ZZZNEWSLETTERSIGNUP1ZZZ
[Tutorials] [Web3D Technology Comparison] [Virtual Humans]
[Virtual Reality] [Art] [People of Web3D]
[Panoramic Imaging] [FAQs] [Companies]