Peter Coffee started out with his round table "Question and Answer" session before dinner. The first question was whether he knew anything about the GIF patent expiring, to which he said "No, not specifically". The main question now is about Linux and where did its code come from. He commented on Microsoft's security posture and of dot-dot files that have code in them. You need a disk to make security updates. He thinks that dot-dot should not be used as a distribution for files. Peter said that 200-250 messages that say some message he sent was rejected as undeliverable that he did not send are appearing in his mailbox so knowing who has sent you a message is more difficult to determine as email addresses are easily harvested. He said that IPV6 would make harvesting addresses more difficult and is taking too long to get out there. IPV6 came out at the wrong time when no one wanted to buy anything that wasn't absolutely essential. The buffer overflow problem we were assured had been resolved 18 months ago has reappeared again. SoBig has been stopped, but still resides on many computers. One problem happening in September is students return to school and hook up with non-secure software on broadband. Somewhere between 25 and 40% of the machines of freshman have been infected according to reports form university administrators. Some sites that you go to won't work when you try to access them with active anti-virus software.

A problem is the prevalence of Internet Explorer that does too much for you automatically. Peter uses Mozilla and he avoids sites that require him to use IE. However, most sites are accessed by IE so they frequently don't worry about any other browsers. You can't look at some sites if you have Active X controls blocked. Peter quoted a comment that if you allow Active X controls it is like accepting a syringe from a stranger and sticking it in your neck. Apple was one site that required it.

Spam is a huge problem. We have yet to find a good spam solution and it is costly. Some companies are willing to accept having all their email routed through another company's server so they can have the spam removed before it gets to them. On the plus side, Beijing blocked 127 servers in China that they identified as spammers. Peter thinks that China wants to create a good impression with the World Trade Organization.

The Apple G5 is a beautiful machine and you could run a virtual Pentium on it. Microsoft never met a standard it didn't like and then Microsoft adds on to the standard and improves it. Everyone else becomes a tributary to the Microsoft stream. Microsoft and Intel have completely compatible positions. XML is ASCII based but Microsoft "ink" complicates things. "Ink" takes the text message and turns it into graphics. XML is not a language, it is a meta-language, a language for writing languages. Peter said that is like being Roman alphabet compatible, but you can still have new language developments like Esperanto. Microsoft is trying to tie XML to its operating system. The next issue of Microsoft will have Digital Rights management tools in its user applications. This will make it difficult for other users to use anything but Microsoft software to use these applications. Management tools have to keep pace with Digital Rights management, but "Rights Management" doesn't work. Every rights management technique has made it inconvenient for legitimate users without ever blocking sophisticated abuse. A perfect example is passwords on Word documents. The main function of a password on a document is to allow you to forget it so you can't use it yourself. You can pay someone something like $100.00 to crack the document for you, and if it is sufficiently valuable you will pay that price. Peter believes the entire process is a waste of time.

What do you think about the Compaq-HP merger? Peter Coffee says they have done far better than he expected. They maintained market share in PCs, but he doesn't know if HP is making money at it. Peter doesn't know anyone who actually regards PCs as a profitable product anymore. Prices have dropped to extremely low levels for high capability systems. Does he think laptops are next? Lap tops still need a higher degree of design and integration than desk top PCs, especially thermal design. Most companies are only buying laptops so they can better utilize office space. Laptops are getting better and better. Hard disks have greatly improved in robustness and price performance. Memory cards are very resistant to shock. You really don't want to walk around carrying a display, a hard drive, or even a keyboard. You can accept limited capability during movement. Put things on a magnetic card and the network keeps track of you. You want to carry something that informs you when you should get interested and gives you a terminal location. You find the terminal and get the detailed information. In the future the laptop may become obsolete like the old heavy portables with CRTs.

Control of crypto by legislation is dangerous. The belief there is some kind of control on crypto that could be useful is wrong. There is a European anti-trust investigation of Microsoft marketing its server products. Is it going anywhere? There are people who are not interested in some future Microsoft products because they believe in two years they will have moved to open source software. Microsoft is fighting open source by claiming sticking with them is the lowest cost approach. Peter says the problem is, he can't necessarily disagree. If the Windows applications are better it will be hard for companies to change. Microsoft was quite successful in fending off OS/2. What does Peter think will happen with the European Union? The threat of losing access to Microsoft technology would put pressure on politicians because processes companies have in place today depend on Microsoft products. People believe they will move to open source software, but they are not ready to do it tomorrow.

The future of display technology, can be stated in one word - "Plastics". Glass is a dinosaur and CRTs take up a great amount of space. CRTs have to go. LCD's are expensive. LEDs are lighter, cheaper and more flexible. CRTs still have better brightness and contrast, but flat panels are improving. Graphics artists are moving to flat panels. They are staying with MACs. If paper was new, it would be considered a great technology. Graphic arts will become much more display centered. The environment will become increasingly display rich and wireless will provide connectivity. You may carry around a very small display for convenience, but you may [not] need it.

At this point, dinner was called and Peter Coffee broke off his "Round Table" discussion.

After dinner (while many people were still eating desserts) Peter began his main presentation. He said it is interesting that everything he said the first time he gave a fall talk to this group has been refuted. He said that he had been quoting Intel people. Intel said it was necessary to move beyond the Pentium architecture because it could never get beyond a 300 megahertz clock rate because the internal instruction set was too complicated. Now inside of the Pentium 4 is a RISC processor surrounded by logic that interprets the old x86 instruction set . The Pentium 4 was embarrassingly good compared to the Itanium. Now Intel has an Itanium 2. The compiler does all the parallelism at compile, which means there are a lot of instructions for information coming across the boundary. There is a lot of cache, 10 megabytes on the central core of the chip. It doesn't do much for your ability to do actual processing. AMD has come up with the AMD64 that does a 64 bit superset of the X86 instruction set. This is an incredibly smart thing to do. The Itanium will start out doing a lot of 32 bit tasks and it will be at a disadvantage compared to the AMD chip. Windows 95 started out by multi-tasking 16 bit applications. Games people and apps get a performance upgrade because they get an immediate 20% boost because they have a dedicated 4 gigabyte playground for themselves instead of sharing it with system processes. 64 bit desk tops will come out soon, particularly games machines. Peter is looking forward to doing comparisons of the new Itanium chip PCs with the Apple G5. The 64 bit processor does improve computing capabilities and works with much larger collections of data. IBM talks about how many gigabytes you can move on and off the chip in a second. Games and nuclear simulations can use this capability to more advantage than floating point performance.

Availability of data to our machines has been greatly improved. Radio transceivers are about where calculators used to be, when you had to pay for them. Now people are giving them away and you find them imbedded in numerous devices. National Semi-Conductor was showing a chip with an embedded transceiver where the transceiver was smaller than a quarter. Intel is pushing a larger version. Peter believes that National Semi-Conductor has the right idea. Progress in this type of equipment was slowed down by SARs. RFI ID tags are spreading quickly, you can have a tag on a carton of soup that tells when it left the factory, where it is now, and the environment it was exposed to. There is an infusion of location information into the entire IT stack. Today Peter expects to get a UPS or FEDEX tracking number emailed to him shortly after his transaction is completed. With RF ID tags assigned to you could get prompt delivery service. You could have personalized mail that gets to you rapidly because the network finds you. Get away from what we usually do with a radio. With radio you have time dimension where you take turns and frequency division. Another way is code division. Peter talked about the way things will be handled in some detail. With good software you can pull out low level signals from data below the noise level. This is being done with consumer products featuring low price and high volume. You will soon be surrounded by an environment that tries to be helpful. What about privacy issues? You should worry about people who have access to your VISA card records that you have now and show your transactions in considerable detail. Privacy has already been compromised.

All of this assumes we can identify people, we can authorize people, and we can hold people accountable. We must have a massive rediscovery of contract law in the digital domain. Now it is difficult to pin down who is who and that the person who has logged on with a password is the person who should have it. Biometrics doesn't work too well. With use of a fingerprint a finger could be removed from someone to provide a finger print. Man in the middle attacks that intercept a request for service to the person who is supposed to provide it can break any of these systems. You can have simulated returns for 802 cards and pick up all kinds of information.

Important current topics are spam and and virtual private networks. Virtual private networks are like locked containers on semis where encrypted material can be sent over open networks such as the Internet. A 128 bit key is fine, but if users are choosing passwords like "mommy" it won't help much. People are enabling system operations without trying to limit people from having bad interactions. Microsoft has always seemed to assume single user systems. Their systems were not designed to have any suspicion that something had come down from some unknown site. There is a gap between the belief of the security you have and the reality. As long as you worry about your security you are probably "OK". If you took a brand new server out of the box and were following Microsoft instructions for downloading a set of updates, on the average you would have multiple infections before the downloads were completed. Companies don't like automated updates, they don't have enough confidence in their configuration management now. They don't want their systems updated in a chaotic manner. If they do use them they prefer to update one central computer and maintain a master configuration, and then update all the rest of their computers at the same time.

All security systems have points of failure in them and when you think you have solved the security problem, you are part of the problem. If you think crypto will solve your problem, you are also part of the problem.

The rings of computing discussed so far are:
1. Processors
2. Connectivity
3. Access the connectivity

The 4th. level is systems that solve other problems or produce problems. Can these systems discover each other and exchange services?

Web services is the next topic. Microsoft discovered the necessity for 3 standards, TCPIT to connect things, ACTP for things to make requests of each other and respond to them, and XML for those requests to have some kind of common language. SOAP is distributed process mechanism that is not blocked by firewalls. Peter thinks they are presenting this as a feature. This is a security problem. A lot of firewalls have exceptions coded into them so the firewall becomes more of a log than a blocking device. Interactivity is indicated by protocol and security still is not implemented properly. The SOAP standard doesn't indicate who is providing the service, there is no guarantee of message integrity.

We can pull out XML data, presenting it, integrating it and rendering it as html. Web services are a great way of producing applications. Microsoft Outlook has access to all of the information on projects and one single data base can hold the information. This provides good unification of project data. Microsoft is doing some great things and rethinking how we package various applications that we thought we knew how to do; and the new ways are more versatile and intelligent.

Questions - Open Source vs. Proprietary Code. Can you do it with open source? Yes you can. Using open source code was central to many plans. Open source code has many people fixing bugs. There is great collaboration from numerous smart people.

Japan, Korea and China are talking about a join effort on a combined Linux platform for software development. Any successful strategy that American or European companies follow requires knowing what China is doing. Intel is building Pentiums in China and designing them in India. Chinese are now producing new proto-type designs of new chips of their own. There is a question about whether the Chinese are putting their improvements back into the Linux information pool.

In Ring 4 or 5, the technology is understood but non-technological things are the problems. People need to take responsibility for things like security and servers sending out spam. The World Trade Organization could help things by setting standards for countries being good trade partners.

RFI tags can be used to mark frequently stolen items, report when they are stolen, and report where they are. RFI and GPS might control a car to stay within a speed limit. Barnes & Noble stopped selling ebooks because of lack of demand. People want to share books and ebooks don't allow this. An attitude could be that individuals buy intellectual property and can't share it, including things like books. Peter believes that the technical community should take an interest in this and try to influence these trends to allow less interference from digital rights laws. Companies have made devices that don't work except with their own products.

We are used to the idea that things we do with things recognized as computers are important. Some applications developers have found fall-out applications of more general use than the specific problem they solved. Processors are much wider spread than the usual computing devices. Your drier could give information to the TV screens and give it an information page for the drier.

Peter concluded his talk and moved on to a question and answer session.

Master Keys, in general, are easy to break. Poorly implemented systems can be astonishingly revealing. Time to resolve something can give information on the key. Peter Coffee says worms that delete files are not the worst thing, "What about worms that just alter files?" This could be much worse.

People have said that most productivity gains could accurately be called the "Wal-Mart Effect". Wal-Mart's demands for quality have raised the capabilities of systems across the board.

A secure system will work if it allows you to track down and prosecute the penetrator when it is penetrated.

There is no secure software, only software-hardware systems can be made to work. You can't install software on arbitrary hardware at an arbitrary site and expect the system to be secure.

Rights management is more about greed than any idea of legitimate ownership.

Wearable computers are great and appliances that communicate are coming.

This was another of the superb presentations by Peter Coffee that he has provided to us year after year. Once again, this write-up is nowhere near like "being there" and getting his witty and highly informative take on the computing world of today and where he expects it to be tomorrow. Peter gives a very large amount of information in a short period of time and if his talk was transcribed it would be about 3 times as long. In many places Peter gave a very detailed discussion that you missed if you were not there.

This was the first meeting of the LA Chapter year and was attended by about 90 persons. This has been our best attendance in years. Much of this can be attributed to AITP and their members who outnumbered the LA ACM members, but even so we had a much better attendance from ACM members than at our normal meetings. No question about it, Peter Coffee draws a good crowd.
Mike Walsh, LA ACM Secretary