Below are items from other talks... use if we find a way...

[ piont here about unicode and websites... too garbled to decipher.. worth keeping?]

BF1

How the Internet accidentally happened.

In class we studied <http://en.wikipedia.org/wiki/ALOHANET>ALOHAnet.

you know Mike Hammer? Reengineering. He knew Dan Bricklin from ? back in Israel. He and I had a seminar in some topic.. Anyway, we studied Alohanet. Very simple . asymmetric. radio one way, dial up the oher. no idea that a line was guaranteed. failure was assumed.

so metcalfe and some other people I knew put it on co-ax. and that was a class project. I sat next to ?Bob. We got 2.99 megabits. and we did some math to show that the throughput estimate was low. but who cares? the real idea was to see what you could do with it. what is it you're measuring against? what's important here is reliable connections ofver an unreliable medium, with nobody making promises. that proved out ideas that would be an Aha for the internet, as opposwed to the arpanet.

What we found was that putting a reliabhle network in the middle acftually became difficult. circuits were expensive and slow.

Using an unreliable transport solved many problems of designing for interconnecting lans. When you're crossing all these protocols, you can say, Well, we'll just ? around it. this ws a powerful idea.

We used BCPL, basic CPL, a probgramming language, very simple, with Multics, as opposed to the very complex PL1, which was mostly a commercial language.. Bhy the way, our PL1 with Multics compiled much faster than IBM's PL1. If you mash BCPL with PL1, you get C. That's what Dennis Ritchie did.

We were plaing with multitasking, and multithreading to make Multics. From our perspective, they cut a lot of corners to make Unix. It was a good tech bed for ideas, even though at first it wasn't up to tne kind of computing that multics was.

On the selling side, the Honeywell salee people made a machine muchmore expensive and complicated. If they hadn't, history would have been much different.

DS: Why Honeywell?

BF: They bought the GE Multics group. And they proceeded to completelymiss the point of multics. And that's why Multics is now of historic but not practical importance.

The problem was that Honeywell made money from memory, and chaged at the base price rather than the marginal incfremental price. Made it more expensive, when it should have been cheaper. Also Multics was made to run on a simple machine, and honeywell made theirs complicated.

BF10

maximize opportunity by focusing on all these things that might work and bet that if you do enough, something will.

DS: How do yoiu move from home networking to internetworking outside the home?

BF: Composited connectivitt from the edge. We composite the roads, right? You have to recognize that when you buy so-called broadband, you do financing in a way that you can never repay the debt. So you never get a chance to say "W've got this abundance. We paid for it. Now we've got these bundles of fibers and we'd like to use them.

You pay $50 a month and after a few months or years you've paid for the cost of installing the fiber to your house. But we keep paying on a loan that never gets paid off. We spend all this money for interenet, TV and phone, and never stop spending it. That's because we live in a cmpany town. It's like that old song "Sixteen Tons": you get another day older and deeper in debt.

DS: What do you think of Net Neutrality and other legislative efforts?

BF: If you're looking for legislation you're asking the wrong question. Net Neutrality is a syjmptom. We should address the basc problem. Which is that we've handed over powers that operate at cross purposes with our need to communicate.

you've got to understand where you get moore's law, and where you get hypergrowth. You can't just take some of Adam Smith and say the free market is going to take care of everythingt.

From emails, to adapt/use in here somewhere...

As to cell/land all I can say is that the discussion of gaming the rules is a reminder that the Regulatorium is a helluvalota fun and consumes huge amount of resources while preventing the creation of new value by putting the entire focus on artificial distinctions. It does succeed in meeting the goal of assuring that abundance does not threaten the incumbency.

As to functional or structural separation -- how does it differ from ATT Divesture? The retail/wholesale division seems to mirror that model rather addressing the fundamental dysfunction in packing up our infrastructure and recycling it as arbitrarily priced services. "Arbitrarily" means that the prices for the bits are set according to a presumed context and a presumed technology.

While I argue that prices are stories I also argue that the so-called costs are -- these are transfer prices set according to arbitrary models and are not market-driven costs. The problem is that the Internet's end-to-end model means that you don't depend on providers' service and they don't control the entire network. Thus they can't increase the price based on such favors. I argue that this was the lesson of Genuity (though Scott Marcus may disagree having been there). What makes the discussion difficult is that it's easy to project are naïve models on the edge where the carriers define reality and thus we worry about doling out scarcity when they could just as easily provide abundance.

The basic problem is that the network has no intrinsic value - all the value comes from how we use the network. If the network had a high intrinsic cost then we might have to deal with that but the glass and copper and radios are so cheap compared with sidewalks and just about anything else that we seem determined to do ourselves harm by pretending it is expensive and then proving this by making it expense in order to ensure that we can meter every bit.

What is the ratio of bits involved in sending an SMS message vs billing for it?

the model of a "hog" holding up all traffic for a long time

is wrong. We're not talking about slow networks or 200 car freight trains

but lots of individual pack vying for slots.

The real problem is with the metaphor of a fixed size buffer -- this is a

Malthusian focus on zero-sum scarcity that creates no new value. In practice

the carriers have given us only fraction of the potential capacity. Remember

the "modem" scare which was 100% correct in noting that if we all started

using modems the phone system would collapse. It didn't because we shifted

to getting access to the inherent abundance available in the transport --

instead of 56kbps per copper pair we could get megabits and the rights of

way now can give us gigabits at lower cost.

What does it mean to connect to an "ISP" if we do our own networking and who

defines fairness? Faux-ATT which, according to the NYT, is now threatening

to judge whether our bits are moral enough, AKA, violate some copyright?

As I said in the longer response -- a good engineer works within the

constraints given -- a great engineer questions the constraints and gets

fired because the constraints serve a policy need higher than mere science

or reality.

Q: What do you thinkn of the noiton of a "Bandwidth Cap: and metered pricing?

What makes this fascinating is the interplay between accidental properties and explicit policy. When you are off-hook you are tying up facilities. It's easy to understand that model -- the story makes sense.

Flat rate may seem as a more efficient billing scheme but it changes your perception of what you are buying. You also no longer have to think of whether the call is worth it.

This is one reason why micropayments schemes have had so many problems catching on -- knowing your being charged creates unease even if it isn't that much. You still think about "is it worth it" because we tend to think in relative not absolute terms.

Flat rate introduces an ambiguity because there is no distinction between the service and the fixed cost connection to do with as you wish.

The ambiguity flips with IP and open VoIP -- the carriers can continue to think of themselves as service providers whereas the users/purchasers can view them as facilities providers.

No wonder it's difficult to talk about these issues. In a way Network Neutrality is at the flashpoint of this discussion. A service provider is in the position of adding value by defining the experience yet those who've assumed a facilities model can't understand why they are meddling.

To the carriers, the future will be the past entrenched.

we agree on the power of competitive markets and the desirability. We need to separate out a stable marketplace model from how we get there.

Perhaps the crucial disagreement is that you seem to think we get low prices because of regulators? I find that strange. If you look at the consent decree that forced IBM to separate the software from hardware business and the way it drove Moore’s law (even if doesn’t’ recognize it’s about economics not physics) you’ll see that those price went down by orders of magnitude whereas the regulated prices stayed essentially stagnant for decades. We saw this even more dramatically in networking where it was able to catchup to the marketplace as within our homes.

The networks within our homes are far closer to the first SQUARE mile connectivity than the carriers’ delivery pipes. Remember that when they owned the wires nothing improved and now that Verizon’s Mocalliance is trying to take control back they are rolling back from flexible gigabit Ethernet to their 100Mbps over coax even when their own VoD just uses 20Mbps. This is sheer incompetence it would be hard to do worse.

If we recognized that the scarcity that justifies telecom policy is false and recognized the inherent conflict of interest then there is a simple solution. DDT – Don’t do that – don’t try so hard to maintain a telecom industry. Instead of local ownership of the local facilities – nonexclusive ownership and management to maximize capacity.

What are these high costs you allude to – if it’s $1K/home for fiber (as I point BAML (Before Applying Moore’s Law)) that’s the cheapest infrastructure we have (OK, maybe something is cheaper but what?). And that fiber is many times what we have now.

But even if you argue about those costs what about the cost of not being able to assume connectivity. The price is beyond money – it’s in lives and quality of life and the inability to take advantage of the abundance inherent in the air, the copper and the glass.

The key is to decouple the infrastructure marketplace from the services marketplace. The actually networking is easy – you just know too much and know it’s hard. That’s why my naiveté gave us home networking. I didn’t see what was so hard about moving packets around – and I still don’t.

As I keep pointing out – the industry is dedicated to preventing abundance – <http://www.frankston.com/?Name=AssuringScarcity>http://www.frankston.com/?Name=AssuringScarcity? Why are we aiding and abetting such an approach? What problem is telecom solving that it’s worth sucking all the value out of our economy just to maintain nonviable businesses.

ON WIRED AND WIRELESS...

The problem with wireless is far deeper than anything that can be fixedwithin "wireless". The fundamental problem is the distinction between "wired" and "wireless" bits.

What would the "wireless" capacity be if we took advantage of the nearly 100% coverage of existing 802.11 access points. For the thought experiment we can assume better protocols than 802.11 including assuming a peer protocol such that phone can a communicate directly without any access point. The term access point itself is biased - it's more a relay point that can connected wired/unwired or relay other ways.

This would open up many many times today's cellular capacity without having any additional cost. Today's cellular does use similar packet ideas but keeps it contained to an infinitesimal portion of the potential capacity.

Why spend time arguing over how long the chain is when we should be discussing freedom?

I agree that "cellular carterphone" is a useful way to get people upset but it's not more than that since the real problem is a cellular architecture designed to assure the incumbency total control. If you pay $50 for a Femto cell you still need to connect into the cellular companies' system.

This is very very reminiscent of X.400 - the failed carrier-defined email system. The only difference is that this time they get away with it. Can anyone on that mole hill in DC understand how very very terrible the entire cellular system is? Can they understand that it's about assuring scarcity

ONE LINERS

There’s a fine line between reciprocity and corruption … if there is a line at all.

There is no intrinsic relationship between the bits and the content. Triple plays are a funding mechanism.

ON THE LIMITS OF GIANTISM

Yes, this It's not a matter of being suspicious of massive projects as being aware of the inherent inability of such projects to work—they should only be deployed for very very very well understand projects such as building yet another bridge and even there it's risky.

ON AT&T vs. VERIZON

In the short term Verizon can gain because of many impediments to connectivity but these are transient impediments. It is much like the modem “crisis” in the mid 90’s when dial up was big. It melted away because it was essentially a configuration error – taking a low rate data stream but locking down circuits and equipment for no added value.

Ultimately they are like ATT in that they have to charge for the value of what people do with the net on the assumption the value is in the network when, in reality, the value is all in how we use networking. Capturing the value in the network comes at a terrible price to the economy as well as our safety. For public policy why do we have an agency – the FCC – dedicated to starving the economy sucking more value out of it that Bernacke could ever hope to add back?

Verizon does its VoD on 20Mbps – so what’s the rest of the FiOS for? The number of bytes is an arbitrary measure because it has a startlingly low correlation with “experience”. We send Flash-created content as video which could be a 1,000:1 ratio (or even million to one) that may appear the same to the viewer. Just look at the quality you can get at <http://www.abc.com>http://www.abc.com if you have your own fast CPU. The connection speed is only one factor and cleverness can compensate. I can give many examples of such anomalies.

I think I’ve sent you the pointer to <http://www.frankston.com/?name=AssuringScarcity>http://www.frankston.com/?name=AssuringScarcity. Perhaps my skepticism is fueled by having started out on Wall St in the 60’s supplying data and being appalled at how little information the traders really used in making their decision – something that Nassim Talib wrote about years later. It’s the astrology dynamic.

What is most frustrating to me, however, is the lack of awareness that broadband has so little to do with the Internet per se. The inability to tap into the already existing coverage from access points is a tragedy – think of all the value created and even lives saved if we can assume IP connectivity.  As I noted at <http://www.netneutrality2008.org>http://www.netneutrality2008.org (for now <http://www.frankston.com/public/talks/SFUNetNeutrality.htm>http://www.frankston.com/public/talks/SFUNetNeutrality.htm though I haven’t linked to it) what if Google put their money into opening up access points rather than the 700Mhz auction? Combine that with protocols that do meshing outside the net (as with Skype) and owners of the pipes lose control.

I’ve used 9/11 and Katrina as examples but the recent Tornado makes me think of the value of micro-weather stations that could easily be deployed to give information and if watches were connected.

What happens when people understand that for $1k (or soon far less) they should be able to get the full capacity of a fiber to their home and not have to pay $50/month for a fraction of that? Perhaps if there were a feature story on how today’s Internet prototype actually works and the possibilities for real connectivity …

ON THE DYING DEAD

This is WAP – the billable wireless internet all over again.

It’s another reminder that the carriers are the dying dead. People want to watch what they want when they want it. The carriers in their deepest bowels want everything to be 1950’s television. Even the TV networks aren’t this dumb and are downloading to PCs or wherever in whatever form people want. The carriers have nothing at all of value – they can only squeeze their orifices and charge for passing through them.

There are simply too many other options – it’s not bypass. That’s a dumb idea. It’s simply exercising choice and people having other options.

As to free, ad-supported Wi-Fi. That’s stupid – that’s like ad-supported lumber. People confuse the Web with the Internet. It’s infrastructure and raw materials. You can advertise to eyeballs for a site but your emergency bracelet must be able to call for help without you having to go to sign in screen and read an add and click to acknowledge that your read the ad or select another billing option. Even if you aren’t ill when you start, you’ll be in shock by the time you get through the process.

It’s copper, radios, glass and nothing more. The rest you create.

When we get past NN we can then get back to getting infrastructure.

AN ANGLE ON VOIP

My POV starts with VoIP so I see things differently. They are taking what taking costs them absolutely nothing and creating billing events. Neat trick.

If you use the WiFi phone within your house to speak to someone in another room does Orange get a cut? If you use it while traveling in the US to call someone else in the US do they get a cut?

Why not just get an open phone and do the VoIP yourself with a cellular path as backup that you hope to rarely use? Skype already does this but it needs to go one step further.

My HP-HW6945 is approaching an interesting threshold. Skype and VoIP apps should take advantage of the ability to have a persistent public IP address (you have to request it from T-Mobile and they’ll give it to you at no extra cost and it’s the norm with Verizon, dunno about Sprint and Cingular. If the voice apps simply maintained an open listen and a very lean listener and automatically take advantage of opportunistic Wi-Fi connections then I would be far more apt to use the IP path.

The big problem is battery life vs all those radios – but that will be solved just as the color laptops went from impossible to the norm years ago. In meantime an app can listen on the cellular side and then attempt to make a Wi-Fi connection. For GSM this can work using CDS (Cellular Data Services) rather than GPRS or other data protocols since you need few bits to rendezvous and then you needn’t pay for the data service. If you are paying for the data service you can the Edge and EVDO path albeit imperfectly but without any additional charge for cellular – even better than a local cell number and you get the world.

ABOUT QOS

My bits are more important than yours and a good QoS system will give me priority because I’m a better person. A bad QoS system would treat us as equals. At least that’s my definition of good.