Groklaw's Sean Daly had an opportunity to meet Vint Cerf, Vice President and Chief Internet Evangelist at Google, at OpenForum Europe last week. Mr. Cerf, known as the Father of the Internet because of being the co-designer with Robert Kahn of TCP/IP protocols and the basic architecture of the Internet, was gracious enough to answer some email questions Sean propounded regarding the future of the Internet, standards in general, and OOXML in particular. Like many others this week, Cerf has been giving the standards process considerable thought, and he concludes in connection with OOXML that "Internet users deserve better
handling of global Internet standards."
Interview with Vint Cerf
~ by Sean Daly
Q: You 've been called the "father of the Internet"... how is your
'baby' doing? What are the trends that interest and concern you most?
I'm just one of the fathers of the Internet! The net serves over 1.3
billion users today, so I am pretty happy about that. On the other
hand we have about 5.2 billion more users still to put online, and
that's a challenge. Increasingly, many users will have their first
access to the Internet by way of their data-enabled mobiles - so it
is a challenge to fashion interfaces to current services that work
well in the mobile context. We still do not have as much broadband
service as I would like to see, so this is another area of interest,
if not concern.
There continues to be a good deal of political
sensitivity to the Internet and the interactions it facilitates. Some
countries understandably worry about abusive content and predatory
behavior (fraud, por nography) and seek to combat it. On the
other hand, there are also attempts to foreclose open debate and
criticism of a political nature, and this worries me. Some of the
mechanisms necessary to the operation of the Internet should not be
politicized. The operation of ICANN (the Internet Corporation for
Assigned Names and Numbers) is one example. The Internet is global
but cultures and social customs vary from one jurisdiction to
another, so the Internet and its operators are finding it necessary
to adapt to these variations.
I am very excited about the maturing effort to incorporate
internationalized domain names (IDNs) into the Domain Name System.
These domain names are made up of strings of characters rendered in
scripts other than the traditional Latin script that has been in used
for domain names in the past. I am equally excited that the Internet
community, especially the operators and software providers, are
responding to the important need for IPv6 introduction into Internet
operation. We need the additional address space to keep expanding the
Internet to serve the world's needs.
Q: Your title at Google is Chief Internet Evangelist, so what is on
the current to-do list?
Cerf: Find ways to increase the implementation of Internet access through
fixed, mobile, and satellite communication channels.
pressing to get IDNs and IPv6 into operation.
Focus on security and
defense against abusive behaviors.
Find more ways to protect user
privacy and to defend against viruses, worms and denial of service
Encourage experimentation with many new applications,
collaborative tools and services, continuing indexing of the World
Wide Web and retrospective indexing of printed material or other
content not yet in digital form.
Q: What were your general impressions of the OpenForum Europe meeting?
What did you take away from it?
Cerf:I thought this was a very timely and well-attended event. I resonated
strongly with its declaration of principles and appreciated the
spotlight it placed on the purpose of standards in the Internet arena
and the processes by which they are developed. I came away persuaded
that the current consideration of the OO XML proposal deserved some
very serious thought. The specification is enormous (6000 pages) and
as far as I am aware, there are not two or more independent
implementations. It does not strike me as a very reasonable candidate
for the Fast Track process that ought to be reserved for standards
proposals that have multiple, independent, interoperable
implementations and stability. I have trouble reconciling "stable"
with 2000 pages of revision proposals that are being considered this
week, again on a Fast Track basis. Internet users deserve better
handling of global Internet standards.
Q: Concerning standards, do you feel that there is sufficient
transparency in the ISO process? Is the secret closed-door policy for
OOXML the exception, or the rule?
Cerf: Terms like "secret" and "closed door" are somewhat "loaded" but I
think any important standards process should be as transparent as
possible. The recent study on open standards underwritten by the
European Commission produced 10 important properties of open
standards that I found very thoughtful. I'm still skeptical of the
adoption of standards subject to intellectual property constraints
and prefer standards that have no barriers to implementation, so
that's an area in which I am not completely aligned with the
Q: Years ago, the open Internet seemed threatened by walled gardens:
Compuserve, Prodigy, AOL of course, and Microsoft Network in its early
versions. Today, some worry that Microsoft in particular is pushing
certain technologies, like Sharepoint, to try to create a kind of
Internet monopoly. Do you share that worry, that the Internet will
break apart into proprietary walled gardens?
Cerf: We saw that walled gardens are NOT what users want. They want freedom
to interact with everyone in convenient and standards-compliant ways.
I do not think we will see walled gardens of the previous kind, but I
do worry when global standards are adopted that are likely to be
implementable by only one vendor. When global standards processes are
overly influenced by proprietary interests, they cease to facilitate
interoperability and competitive implementation. I do worry when
standards are adopted that have potential encumbrances or that erode
the openness that has been a hallmark of the Internet since its origins.
Q: Another issue today is proprietary and patent-encumbered formats.
YouTube is a prime example -- there is no support for unencumbered video
formats like Ogg Theora. Why is it so hard for people to understand
the value of openness?
Cerf: My guess is that some of this arises from concerns for proprietary re-purposing of content without permission. I think a refinement might
be to allow uploading in different formats, at the preference of the
content supplier. There is still the problem that a supplier of
content may not actually have the rights to supply it.
Q: You've mentioned that video is the future and that the Internet is
more than the World Wide Web. Today, we have ISPs and experts claiming
that video is what makes net neutrality no longer an option, because
it will greatly increase usage and put a strain on the Internet. You
cited P2P protocols as a positive innovation, yet the old guard did
and does all it can to destroy P2P, when it would seem that it would
help so much with the problem they posit as solvable only by
throttling the Internet and charging extra for file sharing. That
seems counterintuitive. In the UK, new legislation will pretty much
require ISPs to read every packet to make sure no copyright
infringement is happening. To a simple soul, it seems the "experts"
and "powers that be" and certainly Hollywood content owners have
absolutely no idea how the Internet works or why it does. Perhaps they
are stubbornly clueless, or old-fashioned, or afraid of change, but
doesn't the possibility exist that they could needlessly hinder or
even destroy the Internet before they finally comprehend that what
they are doing is damaging them also?
Cerf: Yes, that possibility exists and the apparent requirement to do deep
packet inspection will only increase the desire by Internet users to
encrypt everything they send for privacy reasons. The behavior of old-
form video providers, focusing on streaming, is narrow-minded at
best. The future of video is file down and uploading at speeds faster
than you can watch (or sometimes, slower than you would want to
watch). In their zeal to deal with copyright infringement,
intellectual property interests could well destroy the openness that
has made the Internet engine run so well.
Q: Some are concerned that Google keeps activity logs on hand for many
months and could mine personal data. Is it really necessary for Google
to hold on to such data at all? Are these fears justified?
Cerf: We have announced that we will anonymize the information after 18
months. The basic purpose for keeping the logs is not to track
personal activity but to lump together related information
transactions so we can refine and improve our ability to respond to
searches in more relevant ways and also to improve our ability to
serve up relevant advertising information when that information is
responsive, e.g., to a search or to a web page on which our ads
appear. You might note that some legal jurisdictions require that
information be retained for law enforcement purposes for certain
periods of time. Telephone records must be retained for periods of
months to years. I think this is also true of email (but you might
want to look more deeply into that). There is a tension, as always,
between protection of privacy and protection of the society from
criminals and finding the balance is not easy.
Q: The distributed architecture of the Internet has been key to its
success It is quite reliable despite massive growth these past 15
years. Yet, a series of cut cables recently limited Internet access in
parts of the Middle East, North Africa, and Southeast Asia. And last
weekend, we saw how a government (in this case, Pakistan) can, under
the right conditions, create a snowball effect and bring sites offline
outside its borders. How can such disruptions be prevented in the
They cannot absolutely be prevented. The problem with major cable
cuts is simply that there are only a finite number of cables.
However, more cables are being laid, but they often have to cross
areas of tectonic activity. Satellite back up can help but is usually
has less capacity.
Richer topological connectivity, avoiding
clustering of capacity in the same physical place all help.
a disaster a few years back in a tunnel in Baltimore, Maryland, where
all the major communication companies put fiber through the tunnel
and a huge fire destroyed a great deal of connectivity at once. As to
the situation in Pakistan, the method used to effect local (national)
access control also had global effects.
Much of the Internet is based
on trust that participants don't lie about networks they are
connected to. There is a secured routing protocol that has been
developed, but it is costly to implement in terms of resources (Secure
Border Gateway Protocol), so there is still debate whether this is a
practical way to remove this particular kind of vulnerability. The
Internet engineering community generally is on the lookout for ways
to improve the robustness of the system and these incidents motivate
Q: If you had all the power in the universe, what would you do next
with the Internet?
1. Get IPv6 running everywhere.
2. Finalize Internationalized Domain Name standards and start
3. Expand edge capacity to at least 1 Gb/s to all termination points
on the network.
4. Get every one of the 6.5 B people on the planet online either by
Mobile or fixed lines or both.
5. Standardize the Delay and Disruption Tolerant Network protocol
suite (DTN), standardize in the Consultative Committee on Space Data
Systems and start deployment for support of manned and robotic space
6. Complete research on semantic web concepts and deploy it.
7. Make speech a first class data citizen for purposes of searching
8. Secure the Internet to defend against a wide range of abuse,
increasing confidence in privacy and security of information and
and that's just for starters...