CFP Fall 2013 Plenary Notes What’s the Internet for, Anyway?

advertisement
CFP Fall 2013 Plenary Notes
By Natalie Klym
What’s the Internet for, Anyway?
Dave Clark, MIT
When the Internet backbone went commercial, it went from research tool to a platform for
(almost) everything. The design goal of the Internet was thus generality, not optimization for
any one application. People kept asking, shouldn’t you optimize for the application of the
day? First it was email, then the Web, and today, it’s streaming video. Given that more than
50% of peak-time traffic is video streaming, primarily from YouTube and Netflix, the lower
layers are being engineered for this very high-capacity type of usage. But is that what we
should be doing, given that capacity is not free?
The answer depends to a large extent on what users value. (The economic or commercial
value generated by the various applications is a different issue, which we will discuss later.)
Even in the pre-commercial days, what users valued was most important; they are at the
center of the universe. And what they want is to connect with each other. Initially this was
through communication services like email, but now it’s primarily through social media like
Facebook.
On a more abstract level, what users value can be thought of in terms of things like
availability and reach, ease of use, trustworthiness, and performance (or capacity). Note that
although the focus has been on bandwidth (capacity), it is not the most important quality.
One way to highlight some of the questions around use value, particularly in terms of its
more abstract dimensions, is to look at rural broadband. For underserved communities,
availability and reach—having broadband at all—are far more important than having it fast.
It’s also interesting to look at individuals who have chosen not to subscribe to broadband,
despite its availability. These people simply see no value, or it’s not something they can
afford, or they are too concerned with privacy and security. But mostly, it’s a lack of any real
practical value.
The question of use value can also be viewed in terms of commercial vs non-commercial
uses. We must not lose track of important public and social uses of the Internet. These uses
justify public sector investment of the infrastructure since the creator of a public good cannot
fully appropriate the value. But here, if we return to my earlier statements regarding the
dominance of streaming video, this high-capacity commercial application may end up being
our best friend because an Internet engineered (technologically and economically) for the
video content industry means it’s suitable for lower-capacity and/or non-commercial
applications by default. In other words, if you engineer a road for tanks, regular cars will have
no problem.
There’s an interesting argument then as to whether or not we need government intervention
to preserve the generality of the Internet as it relates to the technical character of the
platform and the cost of developing non-commercial applications (e.g., Wikipedia) or those
where the economic value is uncertain. [NK: On the other hand, there is the argument that a
commercial Internet will by definition be somewhat hostile to non-commercial uses.]
So far, we’ve discussed the value of the Internet to the users, but there’s another perspective
that looks at the value to the complementers; those who sit on top of the Internet. If you go
to Washington for example and ask them what the Internet is for, they’ll say it’s an enabler of
innovation and economic growth, efficiency and cost reduction, as well as things like
surveillance, which users may not necessarily value.
It’s interesting to note that both the people whose focus is on the user experience and those
whose focus is on the Internet’s impact on the broader economy advocate net neutrality. But
net neutrality is not a goal in and of itself, rather, it’s a means to achieve something else. For
example, one argument is that it prevents discrimination against public goods, or it
encourages innovation.
In conclusion, I’m going to suggest that rather than asking, what’s the Internet for? we should
ask what are the barriers to meeting the needs of the user (availability, ease of use, etc.). Who
are the players responsible for each of these challenges? And to what extent are the barriers
economic vs technical or related to coordination and leadership? In other words, you won’t
change the trustworthiness of the Internet by throwing money at the problem, it’s a deeper,
more subtle problem; it’s about rethinking how we do apps, how we do user interfaces. So
we need to think beyond return on investment and focus on these other things.
Panel:
Hannu Flink, NSN: Given the high growth rate of mobile apps and data, networks need to
undergo a fundamental transformation. Our goal is to deliver 1 GB of personalized data per
person, per day. Video trends factor prominently into our vision, including the evolution of
screens (3D, high- and ultra-def), the integration of video in almost all content (not just
“entertainment”); and an increase in the number of sources of video data. We also factor in
the shift from dedicated devices to cloud computing (return of the thin client), and the
billions of connected objects that will characterize the Internet of Things. Our vision thus
centers around mobile networks with a focus on capacity, reduced latency, self-aware
networks, low energy consumption, cloud computing/thin clients, and personalization of the
user experience.
Tessa Sproule, CBC: As a producer and broadcaster, my questions are: what is TV? and more
importantly whose network is it anyway? The Internet is no longer there to serve us, it’s there
for our audiences to tell their stories. Our role as gatekeeper is changing. Our viewers have
become the gatekeepers, the network, the medium, while platforms like Facebook provide the
doorways to the Internet. Our key challenges now are to make audiences discover our
content, and disrupt the echo chamber effect of social media and the overall Internet
recommendation engine.
Sam Chernak, Comcast: The Internet is for everything; from new businesses to social causes
like ending literacy, etc. Video has saturated the user experience. Comcast is adding capacity
to support that collaborative benefit. Everything can piggyback on that huge payload of video
as it saturates the user experience. The Internet is like fire, it’s elemental. We’re in the early
stages of what is a perfect delivery system for content and connections: capital monies flow
to it. New devices emerge every year. It’s easy to use. And it becomes increasing integral to
parts of life that we can’t even imagine today.
Rob Hunter, ESPN: My question is, what’s the Internet not for? The answer is, it’s not for
broadcast television. Yet, video dwarfs everything else, and we at ESPN are doing out part.
This year, we will deliver more than 3,000 simultaneous events (live over cable tv and online).
We will also deliver 5,000 events online that are not available live. Our shared ambition with
our Comcast partners is to double or triple that because our customers are asking for it. But
traditional broadcasting of programs should not be what the Internet is for. North America
has an incredibly efficient capability to do that—cable TV—which makes the 100 megabits
vision pale by comparison. The Internet should not be used for unicast of old movies (e.g.,
Netflix) or meaningless talk shows produced to fill air time.
Dave Clark, MIT: I’d like to come back to that last comment later on today. Those shows
exist for some reason besides just to fill up the cable channels.
Audience discussion:
Marie Jose Montpetit, MIT: For me it’s about community, connectivity, and content. In
Europe they are working towards digital citizenship, i.e., community. But, as social uses (like
digital citizenship) increase around the world, wireless connectivity needs to be both available
and affordable, both at home and abroad. For example, in Canada, WiFi and 3G are very
expensive, and roaming charges are high for foreigners. We need to make sure we are not
building an Internet for the top 1% only.
Bill Lehr, MIT: In response to Hannu’s 1 GB goal (10x increase), what are people’s thoughts
re is this too much? Not enough? Second question is, how much of this traffic will be
offloaded from cellular to WiFi. How do the wireline and wireless carriers think about that?
Rob Hunter, ESPN: to Hannu – re your goal of 1 GB per day, the average adult consumers 35
hours of tv per week. 1 GB of data is about 20 minutes, so it can’t be for consuming
television.
Sam Chernak, Comcast: I agree. We’ve seen this relentless Moore’s law of consumption
growth, Internet usage. That multiple doesn’t change. Re offloading, we believe that the
intrinsic cost/ability of the wireline is better than any air interface, offloading will happen to
us or by us, but it’s a phenomenon that’s happening. It’s part of our strategy. The millions of
gateways are factored into what we expect to see, in some weird way, it’s a reasonable
equilibrium to work towards
Charlie Fine, MIT: Re what’s the Internet for, I would ask, is it the infrastructure for the opiate
for the masses or for enlightenment? The other dichotomy is who’s paying for this? People
who support enlightenment are often governments while those who support opiates are
often businesses, perhaps there’s a bigger market for the opiates. Maybe those who provide
the opiates will pay in part for the enlightenment.
Natalie Klym, MIT: Given more uses and all the diversity – what are the limits to generality?
Not just technically, but more so on a regulatory level. How do we manage conflicts of
interest? All other applications may be able to piggyback onto a network engineered for
video streaming, but how do you reconcile, for example, the copyright enforcement models
proposed by the content industry with the desires for “sharing” among users?
Sam Chernak, Comcast: It’s a great question, not sure if I have the answer. Technologically,
there’s a lot of runway left, but you do get to the point where the laws of the land, the
behaviors need to reach a point of equilibrium. It can take a long time to find that
equilibrium.
Karen Sollins, MIT: We need to consider all the other smaller devices that are not directly
servicing us; they are the biggest users of the Internet, not us. That changes the economics
and raises issues that are not just about how much video I’m consuming.
Dave Clark, MIT: As computer scientists, we worship 2 gods -- performance and generality
and often ignore costs. A lot of the processors in those smaller devices cost about 10 cents,
so you’re not going to buy a $30/month contract for each one of these devices, so there are
some serious issues re how they get hooked up. Are they part of consumer capitalization of
the home WiFi? Or will there be some other network that has broader reach but different
cost point? Where’s it going to come from? What’s the architecture? It may still be the
Internet but if it has 10-cent interfaces it will be a different part of the ecosystem with a very
different cost structure. That would be a good question for Dina Katabi: You’re trying to
optimize spectrum but how do you hook up a 10-cent processor to the Internet?
Let me leave you with a personal story. Last night I was asked by the Mass Audubon society
if it was possible to develop a GPS receiver and radio for attaching to a small breed of owl
for tracking purposes, that weighed no more than 2 grams and could run for about a month.
That’s a whole other cost and functional point!
Download