Project MAC, where we met S at MIT A the Software Arts building where we worked together T and the attic N where VisiCalc was written
Other writings on our personal sites:

RSS Feeds:



Comments from Frankston, Reed, and Friends

Monday, May 09, 2005

BobF at 3:52 PM [url]:

A Failure to Connect

What is connectivity? I've been thinking about it in terms of networking but that's only a small part of the problem. It's really about relationships and operations that involve cooperation or connections between two "things". In DIY, Not Connectivity, I emphasized that it was about marketplace economics rather than just a lower cable TV bill. I wanted to hand out copies of that essay at the IP Policy Summit and got a case study in the real problem.

So I tried to print out the essay ...

It seemed simple enough -- the hotel had a printer in their business center. But I can't just connect to the printer -- I have to take the file and transfer it to that computer and then pay the per-page price. I borrowed a SD dongle and wrote a copy to the flash file system. The hotel's computer accepted the dongle but it was locked down and the web browser refused to load that file. I would've just connected the parallel port to my laptop but my Toshiba M200 (TabletPC) doesn't have an old parallel port.

I eventually gave up and figured I could just print it nearby at a Kinko's (now owned by FedEx). Even better, I could print across the Internet and have them deliver it! I downloaded the app and eventually got it to the point of transferring the file and I called to make ask if they could deliver it. I was told that such requests require 24 hours! So I gave up and went there carrying just my laptop. I knew I could print from one of their printers and they advertised that they had T-Mobile for wireless connectivity -- I'd even complained about their use of T-Mobile rather than just adding an access point to their LAN. I discovered that I'd given them too much credit -- T-Mobile was for "hot spot" access and not for printing!

Since I was there I asked if they'd print it right away even if I sent it over the Internet. I eventually found someone to say yes. Then, after a few reboots because of software not ready for dynamic environments, I managed to get it to print. And then I waited and it didn't arrive. Eventually I discovered my EV-DO connection would abort before sending the file. So I gave up and switched to my T-Mobile account. But it still didn't arrive and, to my knowledge, is still somewhere in the bowels of their mail system. I finally decided to use my Internet connection to place a copy of the file on my home machine with an MHT suffix (merged HTML) and then view it in the browser. I could then print it but without the nice formatted I had wanted. They "Internet" price was $30 but it was $196 to print it locally. Turns out you are supposed to print one page on the big expensive printer and then use a separate copier for additional copies. Of course they are all the same printing mechanism and, if anything, the copier is much fancier. But the pricing model is ? well, I don't know the rational for pricing model. It just seems perverse. I also realized I could've use this Internet trick to print at the hotel. With some more time I probably could've gotten better formatting via their computers but I had already taken far too much time.

Now why couldn't I just print it directly on the Hotel's printer? All they had to do is put it on the Internet and run a (web?) service that sold "printing". Not only would it have saved me time but I would've been able to take advantage of the capabilities of my computer rather than conforming to the limits of a web page. Creating such a service is quite simple. It's amazing that no one has done it.

Of course it's such an obvious idea that you'd have to be completely clueless to see it as non-obvious. But that wouldn't be enough to prevent it from being patented. Such patents are major impediments to innovation. Such annoyance patents dominate when the innovation process can proceed apace with millions of people able to reduce such ideas to practice. It doesn't take much to implement such a printing service but it's only one element in the larger connected system. A marketplace proceeds very rapidly when you decouple elements so you can innovate.

There is a big difference between connections and dependencies though it may not be obvious. We can compare Bluetooth which makes one dependent upon its profiles with 802.11 which simply provides connections and gives the user the opportunity to innovate. You get fast innovation in the 802.11 world because you don't have to depend on getting the scenarios right -- you take advantage of opportunities and create opportunities for others. Patents lock away opportunity and thwart this process. They don't work well in an engineering marketplace in which each application is built of many elements. This does make it hard to make money by simply having the idea and you may be forced to make money on the printing service itself.

Perhaps the bigger barrier is that hotels are not run by technologists. The printer is treated like a device to be rented out rather than seeing printing as a service. Both models make sense but the latter allows for connectivity -- you can just connect to the service rather than deal with the artifice of the particular device.

Stepping back to connectivity ?

While it's easy to see how we could fix this particular problem the real issue is recognizing connectivity as a general concept. When I first wrote my connectivity essay I treated it as a technical problem -- networking as a means of connecting end points. It's about more than just transferring bits, it's about communicating. The network itself just transports the bits without interpretation but the devices and services on the network need to have a shared "understanding" in order to communicate.

I'm using anthropomorphic words like "understand" and "communicate" because that's the vocabulary that is available but let's not confuse them with people. While we can argue about the philosophy of "understanding" we only need an operational definition -- understanding means being able to exchange information for a purpose. If I send "print" to a device, it will print something. It's that simple and that complicated if we drill down on the details of each application such as asking what formats and protocols we can use to print on the particular printer.

It is indeed easy to lose oneself in the details and difficulties of each problem. That's why we need an architectural view. If the printing service only accepts a particular format and I don't have the drivers to produce the format there's an incentive to provide me with the driver or support more formats. We typically take advantage of meeting places -- such as using Postscript to normalize printers. This doesn't preclude us from taking advantage of special capabilities when they are available but it does move the process ahead by providing a way to move the marketplace ahead without waiting for the perfect solution.

While a service like GM's On-Star will connect you with a service center it doesn't give you the ability to make your own connections. GM does take advantage of wireless connectivity but doesn't do it in a away that makes it extensible so we don't get as much value as we would had the connectivity transport been separate from the service itself. They use connectivity but they don't provide it in the car.

Simply providing connectivity adds new value without necessarily adding additional costs. It can even reduce the costs by sharing the transport and encouraging innovation.

In order to update the Magellan navigation system in my car I have to send the whole unit in. Ostia has a navigation program which can download new maps right into my PDA. Imagine if I could assume that my car is connected.

Note that I used the word "download" because people are used to the term. In practice the relationships are symmetric and data flows in both directions. I can supply navigation updates as I drive around or I can use the connection to report my location home if I choose to tell my family where I am. This is not to say that I should be tracked wherever I am even if the technology allows that -- I should be in control of my own connectivity.

It's useful to apply this connectivity model to banking. When you write a check you are simply specifying a transaction -- you aren't actually transferring the money. Why write a check when I can do my banking online? Typically all the bank does is prints the very same check. This seems silly. All that happens is that the check is scanned into another computer and then, with new banking laws, only the image is preserved. The amount itself is typically scanned in using the computer-readable numbers on the check. This would be funny if it weren't so absurd and so expensive. You can often perform transactions by just entering your account number and bank routing number onto a form at a web site. Credit card systems also perform transactions this way. We can do the same thing we do with checks by just providing some bits and we can use cryptography to assure the bits cannot be forged though they can be duplicated. The same is true of a check but by using the electronic form the banks can quickly determine if the bits, which act like a check, have been presented before. This is a system that can be implemented at the edge. You can consider PayPal as an example of this but it's only for certain kind of transactions.

The purpose of doing direct transactions and avoiding rekeying goes beyond efficiency and reliability. It allows us to exchange other information as well. When I buy a book on Amazon I want a copy of the transaction so I can put it into my own list of books I've but. I don't need Amazon to keep the list and don't necessarily want it. When I transfer stocks, perhaps to make a donation, there is no information describing the transaction. I have to call the donee and tell them about it!

The information isn't even connected. Of course there is a price to pay -- anonymity is important both for privacy and to allow ambiguity. If I give you cash you don't need to know if I'm making a purchase on my own behalf or on behalf of someone else.

Connectivity is not simple and we are just at the earliest stages of understanding the possibilities as well as the issues.

Even if we have a seemingly simple problem task connecting a device to a computer we tend to use old models. Instead of using a special bus to connect a device to a single computer why not just make it available on the network and make the connection logically rather than via a physical path. We're starting to do it with printers so they can be shared but we need to go further and separate the relationships from the accidents of the transport. Of course we also need do assure that we don't' make things too confusing -- you still want the simplicity that comes from using a wire to define the relationship.

I'm using "connection" in a very general way but want to be careful. We can find connections everywhere. I just read that Houdini's cousin was married to Moe of the Three Stooges but I want to focus on connectivity in a technical sense -- the ability to connect devices and other end points so we can establish operational relationships.

While we might be able to reduce our Cable TV bill by using IP-based transport utilities that's a relatively minor application compared with the high value opportunities like being able to print a document at a hotel. Simple but when you need to do it it can be vital. We'll still want to share video streams but the connected video system is about more than viewing programs -- it allows us to create and share our own video streams. It also means that we can get beyond the television grid (even TiVo is only incrementally better) and connect.

Why just watch when I can participate. Connectivity is about enabling opportunities for new kinds of relationships between device as well as people.

BobF at 2:37 PM [url]:

Spam Evolving?

I notice that the email addresses on my sites have not been harvested in a long time. I generate a new one on each visit and including the IP address as well as the date and time. Spam messages tend to arrive in batches with very similar messages and the made-up names tend to be valid English words without an appreciation for the meaning.

What I have been getting are many messages with seemingly random strings of characters. I no longer send failure messages for invalid addresses since they seem to be used to direct the bounces to third party sites and to confirm that spam addresses are valid.

I interpret these changes as indications that the spam is not a widely distributed phenomenon but comes from a relatively small number of prime sources. Those most sociopathic may be shifting to phishing attacks since that�s where the big bucks are.

Much of this falls into the realm of standard criminal activity. All-the-more reason for us to be cautious about web browsers that do us too many favors implicitly. While we do need better mechanisms to know where to place our trust we must be careful about embedding social policies within the basic mechanism of the network. We�ve already seen this in trying to act as if the DNS was like the trademark system. The real trademark system attempts to address normal human approaches to understanding. With the DNS a simple typo can have serious consequences.

It�s important to understand spam and see it in perspective. It�s too easy to give into our fears and see spam as another reason to be afraid of free speech.

For more, see the Archive.

© Copyright 2002-2008 by Daniel Bricklin, Bob Frankston, and David P. Reed
All Rights Reserved.

Comments to: webmaster at satn.org, danb at satn.org, bobf at satn.org, or dpreed at satn.org.

The weblog part of this web site is authored with Blogger.