Thursday, April 30, 2009


Broadband without Internet ain't worth squat

Broadband without Internet ain't worth squat
by David S. Isenberg
keynote address delivered at
Broadband Properties Summit 4/28/09

We communications professionals risk forgetting why the
networks we build and run are valuable. We forget what we're
connecting to what. We get so close to the ducts and splices
and boxes and protocols that we lose the big picture.

Somewhere in the back of our mind, we know that we're
building something big and new and fundamental. We know, at
some level, there's more than business and economics at

This talk is a 30,000-foot view of why our work is important.
I'm going to argue that the Internet is the main value
creator here - not our ability to digitize everything, not
high speed networking, not massive storage - the Internet.
With this perspective, maybe you'll you go back to work with
a slight attitude adjustment, and maybe one or two concrete
things to do.

In the big picture, We're building interconnectedness. We're
connecting every person on this planet with every other
person. We're creating new ways to share experience. We're
building new ways for buyers to find sellers, for
manufacturers to find raw materials, for innovators to rub up
against new ideas. We're creating a new means to distribute
our small planet's limited resources.

Let's take a step back from the ducts and splices and boxes
and protocols. Let's go on an armchair voyage in the opposite
direction -- to a strange land . . . to right here, right
now, but without the Internet.

In this world we have all the technology of today, but no
Internet Protocol, that is, there's no packet protocol that
all proprietary networks can understand.

In this alternate reality, every form of information can be
digitized, BUT there's not necessarily a connection between
all this information and all the users and services that
might discover it and use it to their advantage.

This was the world envisioned by the movie, The President's
Analyst, where The Phone Company secretly ran the world. It's
from 1967, the same year that Larry Roberts published the
original ArpaNet spec.

Roll Clip

In a world without the Internet, it's not clear that we'd
actually have a thought transducer in our brain. But if we
did, I'd bet we couldn't program it ourselves. I'd bet we
couldn't shut it off. I'd bet we couldn't decide who could
receive its signal and who could not.

What WOULD we have?

We would have super-clear telephony. We'd have cable TV with
lots and lots of channels. We'd have lower op-ex and higher
def. We'd probably have some kind of telephone-to-TV
integration so we could order from Dominos while we watched
Gunsmoke. Our cell phones would make really, really good
phone calls . . . and we'd have another half-dozen bungled
attempts to convince us that picturephones were the next
great leap forward.

Surprisingly, we might not have email. The first generation
of Internet Researchers only discovered human-to-human email
in 1972 - the subsequent growth of "People-to-People"
applications was a big surprise to them. Now, without email,
there there'd be no reason to invent the Blackberry or the
iPhone. Without the Internet, it would be a voice, voice,
voice, voice world.

This voice, voice, voice would be expensive. Without the
Internet - specifically without Voice over IP -- we'd still
be paying fifteen cents a minute for long distance, because
VocalTec would not have commercialized VOIP, Vonage and Skype
wouldn't exist, and even the major telcos would not have used
VOIP to destroy the international settlement system.

Data service? Think ISDN. Actually, think about a dozen
different so-called Integrated Services Networks, each with
its own access and login, with no good way for one to connect
to another. Metcalfe's Law would suggest there'd be orders of
magnitude less traffic overall.

Would we have Search? Perhaps. Imagine what Encyclopedia
Britannica On Line would look like in a non-Wikipedia world .
. . at a buck a lookup.

Digital photography? Perhaps . . . but medium would be paper
and the biggest company would be Kodak.

What about Amazon? EBay? YouTube? Google Maps?
Travelocity? Yahoo Finance? iTunes? Twitter? Facebook?
CraigsList? Blogging? On-Line Banking?

We wouldn't even have Web sites. Sure we could probably buy
some kind of proprietary on-line presence, but it would be so
expensive that only GE, GM and GQ could afford it, and so
inaccessible they probably wouldn't want to pay.

Web 2.0 - the ability of a single computer to reach across
the Internet in a dozen different directions at once to build
an customized web page on the fly - would be worse than
unavailable, it would be unthinkable.

But it's not all bad. Without the Internet, we would still
get our news from newspapers, the corner bookstore would
still be down on the corner, the Post Office would be
thriving, your friendly travel agent would still be booking
your trips, Dan Rather would still be on TV, perverts would
still get their sick pix in inconvenient plain brown
wrappers, and the NSA would not know the books I bought at
Amazon or who I email with.

Tough. We lost a lot of skilled leather-smiths when they
invented the horseless carriage. We'll find ways to deal with
the Internet's changes too.

Without the Internet, the minor improvements in telephony and
TV certainly would not drive the buildout of a whole new
infrastructure. The best way to do telephony would still be
twisted pair. The best way to do Cable TV would be coax.

Now I'm a huge Fiber to the Home enthusiast! But I'm also
part of the Reality Based Community. So let's face it, even
WITH the Internet, including Verizon's amazingly ambitious
FIOS buildout, the business case for fiber is so weak that 97
percent of US homes still aren't on fiber. We are still in
"Law of Small Numbers" territory. The Internet is the only
thing standing between our limited success and abject

Notice, I have not yet, until now, used the word BROADBAND.

But before I talk about broadband, I want to talk about
Synechdoche. Synecdoche is when you say, "The Clock" but you
mean Time. Synecdoche is when you say, "Eyeballs," but you
mean The Customer's Attention. Synecdoche is when you say,
Dallas, but you mean, "The Mavericks."

Most of the time Broadband is synecdoche. When we say,
"Broadband," most of the time we mean, "High Speed
Connections to the Internet."

I repeat, Most of the time when we say Broadband we mean High
Speed Connections to the Internet. Broadband is synecdoche.

Without the Internet, "Broadband" is just another incremental
improvement. It makes telephony and TV better. It makes the
Internet better too. But the key driver of all the killer
apps we know and love is the Internet, not Broadband. And, of
course, the Internet is enabled by lots of technologies -
computers, storage, software, audio compression, video
display technology, AND high-speed wired and wireless

Now, Broadband is a very important enabler. The United States
has slower, more expensive connections to the Internet than
much of the developed world. And that's embarrassing to me as
a US citizen.

Imagine if a quirk of US policy caused us to have dimmer
displays. That would be a quick fix, unless the display
terminal industry demanded that we disable the Internet in
other ways before it gave us brighter displays. Or insisted
"all your screens are belong to us."

High-speed transmission does not, by itself, turn the wheel
of creative destruction so central to the capitalist process.
The Internet does that. Broadband, by itself, does not fuel
the rise of new companies and the destruction of old ones.
The Internet does that. Broadband by itself is not
disruptive; the Internet is.

The Internet derives its disruptive quality from a very
special property: IT IS PUBLIC. The core of the Internet is a
body of simple, public agreements, called RFCs, that specify
the structure of the Internet Protocol packet. These public
agreements don't need to be ratified or officially approved -
they just need to be widely adopted and used.

The Internet's component technologies - routing, storage,
transmission, etc. - can be improved in private. But the
Internet Protocol itself is hurt by private changes, because
its very strength is its public-ness.

Because it is public, device makers, application makers,
content providers and network providers can make stuff that
works together. The result is completely unprecedented;
instead of a special-purpose network - with telephone wires
on telephone poles that connect telephones to telephone
switches, or a cable network that connects TVs to content -
we have the Internet, a network that connects any application
- love letters, music lessons, credit card payments, doctor's
appointments, fantasy games - to any network - wired,
wireless, twisted pair, coax, fiber, wi-fi, 3G, smoke
signals, carrier pigeon, you name it. Automatically, no extra
services needed. It just works.

This allows several emergent miracles.

First, the Internet grows naturally at its edges, without a
master plan. Anybody can connect their own network, as long
as the connection follows the public spec. Anybody with their
own network can improve it -- in private if they wish, as
long as they follow the public agreement that is the
Internet, the result grows the Internet.

Another miracle: The Internet let's us innovate without
asking anybody's permission. Got an idea? Put it on the
Internet, send it to your friends. Maybe they'll send it to
their friends.

Another miracle: It's a market-discovery machine. Text
messaging wasn't new in 1972. What surprised the Internet
Researchers was email's popularity. Today a band that plays
Parisian cafe music can discover its audience in Japan and
Louisiana and Rio.

It's worth summarizing. The miracles of the Internet -
any-app over any infrastructure,
growth without central planning,
innovation without permission,
and market discovery.
If the Internet Protocol lost its public nature, we'd risk
shutting these miracles off.

One of the public agreements about the Internet Protocol lays
out a process for changing the agreements. If somebody
changes their part of the Internet in private, they put the
Internet's miracles at risk. Comcast tried to do that by
blocking BitTorrent. Fortunately, we persuaded Comcast to
stop. If it had continued, it would have put a whole family
of Internet applications at risk, not only for Comcast
Internet customers, but also for everybody who interacts with
Comcast's customers.

The whole fight over Network Neutrality is about preserving
what's valuable about the Internet - its public-ness.

The Internet threatens the telephone business and the cable
TV business. So of course there's a huge propaganda battle
around the Internet.

The propaganda says Network Neutrality is about treating
every packet exactly the same, but the Internet has never
done that. The propaganda says that Network Neutrality is
about regulating the Internet, but we know that the Internet
exists thanks to the government's ArpaNet, and subsequent
wise government regulation.

Look who's calling for regulation anyway! The only reason
telcos and cablecos exist is that there's a whole body of
franchises and tariffs and licenses and FCCs and PUCs keeping
them in business.

Cut through the propaganda. Network Neutrality is about
preserving the public definition of the Internet Protocol,
the structure of the Internet packet, and the way it is
processed. If there are reasons to change the Internet
Protocol, we can do it in public - that's part of the
Internet too.

It's the Internet, smart people. Your property already has
telephone and TV. So does everybody else's. Broadband without
the Internet isn't worth squat. You're building those fast
connections to The Internet.

So please remember that the essence of the Internet is a body
of public agreements. Anti-Network Neutrality attacks on the
public nature of the Internet are attacks on the value of the
infrastructure improvements you've made to your property. So
you can't be neutral on Network Neutrality. Take a stand.

If you install advanced technology that makes your property
more valuable, you deserve your just rewards. But the
potential of the Internet is much, much bigger than your

Like other great Americans on whose shoulders I stand, I have
a dream. In my dream the Internet becomes so capable that I
am able to be with you as intimately as I am right now
without leaving my home in Connecticut.

In my dream the Internet becomes so good that we think of the
people in Accra or Baghdad or Caracas much as we think of the
people of Albuquerque, Boston and Chicago, as "us" not

In my dream, the climate change problem will be solved thanks
to trillions of smart vehicles, heaters and air conditioners
connected to the Internet to mediate real-time auctions for
energy, carbon credits, and transportation facilities.

In my dream, we discover that one of the two billion who live
on less than dollar a day is so smart as to be another
Einstein, that another is so compassionate as to be another
Gandhi, that another is so charismatic as to be another
Mandella . . . and we will can comment on their blog,
subscribe to their flickr stream and follow their twitter

But I also have a nightmare . . .

In my nightmare, the telephone company has convinced us that
it needs to monitor every Internet transaction, so it can --
quote-unquote -- manage -- what it calls "my pipes".

Maybe it says it needs to stop terrorism, or protect the
children, or pay copyright holders. Maybe there's a genuine
emergency -- a pandemic or a nuclear attack or a 9.0

In my nightmare, whatever the excuse -- or the precipitating
real-world event -- once the telephone company gains the
ability to know which apps are generating which packets, it
begins charging more for applications we value more.

In my nightmare, once the telephone company has some
applications that generate more revenues because they're
subject to management -- and others that don't -- the former
get all the newest, shiniest, fastest network upgrades, while
the latter languish in what soon becomes Yesterday's Network.

In my nightmare, new innovations that need the newest fastest
network, but don't yet have a revenue stream, are consigned
to second-class service. Or they're subject to lengthy
engineering studies and other barriers that keep them off the
market. In other words, in my nightmare, all but the most
mundane innovation dies

So it's up to you. When you make high-speed networks part of
your real estate, if you insist that these connect to the
REAL Internet, the un-mediated, un-filtered publicly defined
Internet, you're part of a global miracle that's much bigger
than your property. Please ask yourself what's valuable in
the long run, and act accordingly.

Technorati Tags: , , , ,

"The propaganda says Network Neutrality is about treating
every packet exactly the same, but the Internet has never
done that. "

Indeed it hasn't, but some have argued that it should. Welcome to my side of the debate.
I'm not on Richard Bennett's side of the debate.
Interesting, David. You have here yet another definition of "network neutrality" which is different from anyone else's definition.... Yet more evidence that the term is semantically null. And you seem to equate network management and monitoring with something evil -- rather than something which is necessary to keep the network running. Finally, you equate providing tailored service (e.g. especially low latency for applications that need it) to something evil as well. If the government were to enforce this view, then new applications and innovation would be hindered by the non-availability of such service.
Brett, I am not defining Network Neutrality in my speech, I am saying what it is about. C'mon you can say that a car is about transportation without defining a car. Jeebers H. Crustacean! The basic un-spun definition of NN has to do with non-discrimination on the basis or origin, destination or content. Surely you've seen something like that? And, as you know, I question whether NN, so defined, is sustainable under the propaganda onslaught of the incumbents, and, if you've been listening to me, you know I favor an exemption for small WISPs.
Well, David, if you believe that "network neutrality" has to do with non-discrimination with regard to origin, destination or content, then you would have to believe that Comcast did nothing wrong when it throttled P2P traffic. After all, that throttling was completely content-neutral and merely had to do with behavior -- the user was abusing the network. Right?

Wrong. Comcast sent RST packets based on its analysis of the contents of the packets it wanted to screw with.
Good job, Dave. I'd summarize it as:

"Without Internet, Broadband is, at best, CableTV."

(or PictureFonz - grin)

The whole Network Neutrality issue, I believe, turns on the duplicitous use of the word "Competition".

To the consumer, "competition" means selecting a provider from a variety of suppliers based on real, demonstrable differences in service and price of the offerings, and then being able to hold the provider accountable by easily moving to a new provider if the promises don't match reality.

To the FonKompanies (including CableTV), "competition" is about each of them having the same chance to lift your wallet, because once they get it, they hang on for dear life with the aid of terms, conditions, and early termination fees as they have so clearly demonstrated in their cellular offerings.

If customers could easily vote with their feet, carriers might be more retiscent to anger them en masse. But since that ain't gonna happen any time soon, mandating good behavior would seem appropriate.

And Brett, Network Neutrality is not "semantically null" and you know it. It may suffer from intentional ambiguity, but that hardly makes it "null".

"Content" in the NN debate is about selective, discriminatory behavior based on looking at UDP or TCP packet headers just as much as it is looking at Application Payloads.

One curiosity is that it's clear many (but certainly not ALL) customers WANT certain kinds of controlled discrimination based on Application Payloads - those are called "Spam Filters" and "Anti-Virus Scanners". So there is a place for service providers and customers to agree upon specific intrusive behavior, but it requires full-and-fair disclosure and a clear, demonstrable value to the customer. The bug in the current situation is that the mechanism for a customer earning treatment as a responsible adult is usually poorly thought-out if not down-right missing.

In the case of applications which open multiple streams to game the existing fairness mechanisms, this gets down to the interaction of the fine-grained technical definition of "fairness" and the product-definition notion of "fairness". I believe the current machinery is woefully inadequate because it doesn't have enough information to do the job without descending into ad-hoc tomfoolery which is arguably just as bad in its own right.

Keep swingin', Dave.
Well written. I think it comes down to whether broadband subscribers are allowed to use the Internet or whether they are gated by Comcast et al., who find it profitable to be an annoying intermediary.
You are confusing synecdoche with metonymy -

When A is used to refer to B, it is a synecdoche if A is a component of B and a metonym if A is commonly associated with B but not actually part of its whole.

Therefore, Dallas could be a synecdoche for Texas or a metonym for Mavericks, etc.
You seem to be overvaluing the structure of the packet, and undervaluing end-to-end connectivity. Perhaps not surprisingly, since end-to-end connectivity is fairly broken on the internet these days.

The mere fact that the packet promotes end-to-end connectivity isn't enough, though, as we've seen with IPv6 deployment. That support isn't useful without an infrastructure on which to run it.

So while I don't dispute that you are on to something with your view of what network neutrality is, I think you miss the boat when you don't even mention end-to-end.
Here is one of those issues that is ridiculously important to everybody who uses the internet, but an issue that nevertheless is misunderstood, and to a great many people, completely unheard of.

Your description of what could have been sounds very similar to Minitel [].

Here in the UK, we have been slow to catch up with things such as digital and cable TV. I remember the chin-stroking newspaper columnists debating whether or not we had too much choice when they started to get more popular.

But can you imagine how shitty the internet, and in particular www, would be if it only had 500 channels? The internet very nearly could have been this, had it not have developed and spread so freakishly quickly.

The ability of the net to rise above those attempts to gimp it, regulate it or otherwise ruin it is the best thing about it. It is something worth saving.
Metonymy, synocdoche, let's call the whole thing off :-)

I think I get it. If I say "Obama" to mean, "The current US Executive Branch," that's synecdoche. If I say, "The White House," then arguably that's *not* synecdoche, because the White House is a place, not a member of government.

So if I think of the Internet, tightly defined, as the set of public agreements that define the IP packet and how it is processed, then the "anonymous" is correct, "broadband" is metonymy. But if I think of the Internet as the network of networks, and everything comprising it, then "broadband," as commonly used to mean, "high-speed connections to the Internet," is synecdoche, as is, "my computer," in the phrase, "I'm going to look up that Web site on my computer."

Mike, SMTP is one of the great tragedies of the Internet, because of its susceptibility to spam. You are wrong to imply that spam filters add value, but other filters are anti-neutrality.

If someone were to suggest that every customer of an ISP should have to go through a filter to access web pages, there would be (and has been!) a great hue and cry over it. Why? Because web pages are not forced on you. If a web page contains something offensive, generally speaking you needn't go there, and needn't see the offensive material.

On the other hand, with SMTP, the situation is reversed - content reaches out to *you*. And so you have to have a filter to stop bad content from coming in, because you aren't the one who chooses what content arrives.

The SMTP/Spam/Spam filter problem is an excursion down the slippery slope of non-net-neutrality. It is a problem we need to eliminate through better protocol design, not the exception that proves the rule.
David, you say: "Network Neutrality is about preserving the public definition of the Internet Protocol, the structure of the Internet packet, and the way it is processed."

I'm not aware of anyone on either side of the network neutrality debate who has proposed changing the structure of the IP packet (I assume you mean the IPv4 packet, as IPv6 has a rather different structure.)

Can you explain what you mean by this?

As far as it goes, the IP packet has very little structure and is subject to very little processing. IP is essentially a null protocol, and as a result the heavy lifting in networks is done by the system that carries IP as payload. The real heavy lifting in networks is done by the link layer or by MPLS.
Where's comments 5-10?
Strangely, the "Leave a comment" page shows 13 comments, but the main page only shows 4.

Wordpress is a nice, open source blogware package that doesn't require Google login and shows comments just fine. Check it out.
This is magnificent. If it's possible, better than your "Stupid Network" paper. I'm going to have the old fogies in my life read this, and I think they'll finally get the internet.
Comments fixed?
Great stuff! Thanks for the insight..

Broadband is not the internet, Yes.
The internet's core principle is what you label "public." Yes.

To maybe unpack that public a bit: the internet is public in the sense that it enables more nearly frictionless communication over an expanding number and range of modalities (aka protocols, hardware, software, etc.) One to one (email, voip and on). One to many (all the flavors of broadcast). Many to one (think RSS feed aggregation). It makes communication available synchronously or asynchronously—and both or points inbetween (think email lists to archives).

Before I deginerate entirely into lists ;-): What the internet changes is make unfettered communication easy, reliable, and fast.

The core principle of "public" is, perhaps, unfettered communication.

You remark that the internet includes mechanisms to publicly adjust its workings...that's sounds like the basic function of a good constitution. But maybe we need as well a bill of rights—things that can't be trespassed upon freely by "legislation" enabled by the "constitution." I would nominate "unfettered communication" as the first right/principle listed in such an internet bill of rights.
Inspiring as hell. Thank you.
I hesitate to object due to what I detect is a firm belief that "Broadband without Internet ain't (sic) worth squat" BUT there are times that specific, dedicated "private" applications of broadband might limit or deny internet connectivity. It seems that the mindset of the author is primarily based upon the commercial and, perhaps the social benefits of a holistic communications medium. I certainly support the merits of those beliefs. We ought not assume these benefits are the only potential value to other applications of broadband technologies. As capacity and integrity issues grow, specific communities may find dedicated broadband networks more reliable given the mission of the organization and/or community.
It seems to me that the issue with "heavy hitters" is a non-issue no matter how you slice it.

If you argue that "heavy hitters" are hurting your customers, it's either because
a. your network design is inadequate for internet consumption (as in cable before docsis 3.0)
b. your aggregation network is inadequately provisioned (and this is an issue with backhaul that you have to purchase, but then this suggests that someone is ripping you off or has an inadequate backhaul network)

Once you traffic hits the internet, there are no heavy hitters, it's a self-regulating flow.

As I suggested at F2C, in the fiber world it would be easy technically (although we might be a few years away from this financially) to dedicate a wavelength to internet access, guaranteed "no tampering".

In the meantime, if telcos still use that argument, maybe they can offer a way of guaranteeing internet bandwidth over DSL so that customers who were told they were buying 8 Mbps actually get that? Because if they can't, it seems to me they don't have a leg to stand on.

David, I'll be blogging about this, and would also like to translate it in French, with your approval, of course.
David, is the next question: is there such a thing as 'neutral traffic management' and if so, how should it be defined? I ask on the basis that to be paid to gerrymander the routing to favour some traffic is clearly wrong, but managing the traffic to optimize the whole Web 'experience' might be the way forward.
Dave, you write:

Comcast sent RST packets based on its analysis of the contents of the packets it wanted to screw with.No. It sent RST packets based on the user's behavior, without knowing or caring what the packets contained. In fact, the Associated Press documented the fact that Comcast's management of abusive P2P behavior was neutral, when it observed that Comcast's equipment did exactly the same thing to a transmission of the King James Bible as it did to illegal music and movie uploads. It didn't care a bit what was being sent. This is a good thing, IMHO. ISPs can't be responsible for policing content. But they must maintain order and quality of service on their networks and must stop abuse, or they are not acting responsibly.
Brett Glass says, "It sent RST packets based on the user's behavior, without knowing or caring what the packets contained."

Comcast sent RST packets based on the application protocol being used to convey the data. Specifically, they "managed" the following application protocols in this way: Ares, BitTorrent, eDonkey, FastTrack, and Gnutella. In order to make this distinction, it is necessary to analyse packet payloads to some extent, and Comcast did analyse application-level metadata to some extent (precise details were not disclosed). So it's true to say that "Comcast's management of abusive P2P behavior was neutral" in the sense that it "did exactly the same thing to a transmission of the King James Bible as it did to illegal music and movie uploads", but it was entirely non-neutral about the application protocol whereby those uploads took place.

So if by "based on the user's behavior" you mean "based on the user's choice of application software", then your remarks are true. David Isenberg's remark, that "Comcast sent RST packets based on its analysis of the contents of the packets it wanted to screw with," is also true, and you have not contradicted it, despite the fact that your remarks are phrased as a rebuttal. The disconnect between the two is the result of a line drawn at different places: Dave draws the line of acceptable analysis somewhere around the transport layer headers; you've drawn it somewhere up above the application layer, between the application's distinction of data and metadata. David says they crossed the line, and you say they did not: you're both right, because you've drawn two quite divergent lines, and Comcast's behaviour fell in between them.

See a CircleID article I wrote for more detail and citations about the Comcast practices in question.
Post a Comment

This page is powered by Blogger. Isn't yours?