More Junk Mail from Bob!
Friday, October 29, 1999
Congratulations or condolences on receiving the dozenth Junk Mail! If you are new to the elite and suffering group who receive this questionable piece of propaganda and you want to see the other junk mails for some strange reason, go to Bob's Junkmail
. You can also, sign up your friends or, more likely, your enemies to junk mail there.
Have you ever chatted online? Be one of the first 15,000 people at http://www.townhallmeeting.excite.com
a week from Monday and you can chat with President Bill. (This link doesn't work today. Either they'll get it going by then or I have the wrong link.) I'm wondering how much Excite donated to who for this. I'm also wondering if Clinton will do his own typing and spelling. Hmmm... I bet he gets an intern for that.
Today's Y2K Update is combined with Your Wasted Tax Dollars to bring you: Y2K for Kids, the Web Site.
Why is the General Services Administration doing this? And why do they tell kids, "You may not have electricity for a day or two?" They're trying to scare people when nothing's going to happen!
In other Y2K news, Boston plans to celebrate the New Year at 7:00 pm to allow people to get really soused before the inevitably breakdown of society at midnight.
Big news from October 20: "The Encyclopaedia Britannica will make the entire contents of its 32 volumes available for free on the internet on Wednesday." Nine days later, you can go to Britannica.com and it STILL doesn't work! It is one of the biggest screw-ups I've ever seen on the web. Now when you go to http://www.britannica.com
, you get a nice apologetic message (dated 10-28) from the CEO that mentions, "I now believe we will be able to serve a significant number of users beginning sometime next week." Oops! Well, when they get it going I bet it will be useful. They're putting their entire encyclopedia on the web partly because their sales have been declining.
Today the Dow Jones average went up over a hundred points to 10,731. Seventy years ago today, the stock market crashed. It started down on "Black Thursday" the 24th, but the big crash was on October 29th. The Dow Jones average went from 380 earlier in the year to below 50 (I think). Here's what the AP wrote on October 30:
Last Monday Payne Stewart and some other people died in a Lear Jet that lost pressure. It looks like they passed out after the plane lost pressure because they didn't get their oxygen masks on. The plane was apparently set to climb at a certain rate without the target altitude selected because it climbed really high, and probably stalled and went down several thousand feet, then repeated this sequence until it ran out of fuel.
About a month ago I was taking some remedial (or was it recurrent?) flight training. The instructor pulled back the throttle and said, "The engine just died." I started gliding toward a nearby runway. Then the instructor said, "You just lost pressurization, too, and you're at 29,000 feet." I started thrashing around trying to get the oxygen mask out from under the seat next to me while maintaining at least a little control over the airplane. I never did get the mask out. In all the commotion I couldn't find the button to open the door where the masks are stored. NOW I know where that button is.
Yesterday a Blue Angels jet crashed, killing two Blue Angels pilots who were in it. That is really scary because those guys are a LOT better at flying planes than I am. Here's a picture of our plane with Blue Angels 3 (7 is the one that crashed):
Blue Angel 3
George Washington Ditch (yes, it's real):
George Washington Ditch
Cathy and Melinda in Utah:
and I40 into the storm:
Two weeks ago I mentioned in an earlier junk mail (Junk Mail 10
) that "Neural networks have been researched for a lot of years, but as far as I know they haven't been good enough for much useful application." It turns out that I didn't know very far. John Chandler emailed:
"Artificial neural networks (ANNs) have been useful in many applications. Casimir "Casey" Klimasauskas, founder of Neuralware, Inc., which sells one of the best packages, spoke here a couple of years ago. He said, "Statisticians say that neural networks are nothing but nonlinear regression. That is true, but ANNs are nonlinear regression in a form that can be used by semi-skilled personnel." Klimasauskas wrote his first ANN program, and started his company, on his kitchen table, funded by his VISA card. Interesting guy. Smart; Caltech grad. Reasonably humble.
"One well-known successful early application was predicting the probability that a loan applicant will declare bankruptcy. That was maybe ten years ago.
"The best free ANN packages are probably SNNS, the Stuttgart Neural Network System, and Jetnet, from Sweden (written in FORTRAN 77 by physicists, my kind of software!). Neureka is popular but doesn't look good to me.
"Matlab has a popular ANN module. Definitely not free!
"There is an ANN program that is the best backgammon playing program in the world. Written by Gerald Tesauro and controlled, unfortunately, by IBM."
Chandler's a wayward physicist who's been lost in the OSU Computer Science department for a few years.
In February, they are supposed to have "a working draft of the human genome." "They" refers to U.S. and British labs doing government-funded research. "A working draft" means that they'll have about 90% of the U.S. genes mapped. This is about a year ahead of schedule.
There's a guy named Ventor, however, who is planning to patent my genes sequences. His company, Celera, has applied for 6500 patents on human gene sequences in a single month. Never mind that they have not actually proven anything. These are preliminary patent applications, something the patent office started four years ago. Celera filed these patent applications so they could own the rights to these human gene sequences in case they prove to be correct and useful. To put this in perspective, IBM got 2657 patents in 1998, more than any other company ever. Incidentally, you can now buy and sell patents online at http://www.pl-x.com
I think that human genome should not be patentable. Patents are for ideas and methods, not for information. Celera filed this huge mess of patents saying that it's a new and novel idea to use some particular sequence of genes for medical purposes. The sad part is, the U.S. Patent and Trademark Office normally awards patents on stuff like this. In fact, the Patent office recently instituted on-line patent filings and the first one done, on September 29, was a patent for a gene sequence listing.
But I think we might get lucky in this case. The U.S. and British governments are considering making the human gene sequences public domain information. This would be good. There are, in my opinion, too many stupid patents already being issued.
Amazon.com recently sued Barnes and Noble. Why? Because on Barnes and Noble's web site, they let you save your address and credit card number so you can buy something without re-entering the information. So what? Amazon.com has a patent on this. That is REALLY stupid. There is nothing innovative about saving customer information. They just filed a patent to do it on the internet a couple of years ago, and they got the patent last month on September 28. Here's what it says:
"A method and system for placing an order to purchase an item via the Internet.
"The order is placed by a purchaser at a client system and received by a server system. The server system receives purchaser information including identification of the purchaser, payment information, and shipment information from the client system. The server system then assigns a client identifier to the client system and associates the assigned client identifier with the received purchaser information. The server system sends to the client system the assigned client identifier and an HTML document identifying the item and including an order button. The client system receives and stores the assigned client identifier and receives and displays the HTML document. In response to the selection of the order button, the client system sends to the server system a request to purchase the identified item. The server system receives the request and combines the purchaser information associated with the client identifier of the client system to generate an order to purchase the item in accordance with the billing and shipment information whereby the purchaser effects the ordering of the product by selection of the order button."
If you actually wade through all this obtuse text, you'll notice that there's pretty much zero here that's new and innovative. This sort of thing is pretty common in the software business. People get patents on amazingly simple, non-innovative ideas. Someone who unknowingly has the same idea then has to prove that even the most obvious has been done before the patent was applied for, either that or pay the patent owner a bunch of money. It's expensive and time-consuming to fight a patent.
Companies like IBM, Microsoft, and Intel patent everything they can and then use them as bargaining chips for "cross-patent licensing." This protects them most of time from patent infringement lawsuits. If someone sues them for patent infringement, there's a reasonable chance that the suer is infringing on one of the suees thousands of patents. Nice way to do business, huh?
There are now companies that are in the business of software patent owning. They find a failing software company who has a patent and buy that company at or before the bankruptcy auction. Or they buy the patents from a cash-strapped company for peanuts. They get seemingly worthless patents anywhere they can find them. Then they search high and low for someone infringing on these patents and sue them. This wouldn't happen if the Patent Office didn't award so many dumb patents. Maybe the people in the patent office don't know much about software.
I remember a few years ago at Comdex a lawyer who was walking around to all the exhibits, looking for software products that had the mouse cursor move at a faster rate when the mouse moved faster. His purpose was to extort royalties from these companies, because his company owned a patent on this. I don't know how successful he was, but I do know that this idea had been around for a long time before that.
Another stupid patent is the GIF patent held by Unisys. Unisys, a company who bought a company who bought a company etc., ended up with a patent on the GIF file format. You might ask how someone can patent a file format. Actually, the patent is on the compression algorithm used to read and write GIF files. The compression algorithm, now called Lempel-Ziv-Welch or LZW, was published by IEEE in 1984: "A Technique for High Performance Data Compression," Terry A. Welch, IEEE Computer, June 1984. It is a variation of the Lempel-Ziv compression method that was published a few years before that. It's been in common use ever since. For example, here's a quote from the Proceedings of the 1997 IEEE Data Compression Conference: "The Lempel-Ziv codes are universal variable-to-fixed length codes that have become virtually standard in practical lossless data compression." (This paper was on LZW compression and made no reference to Unisys.)
A virtual standard? How can that be, if Unisys has the patent? Actually, I'm not sure, especially since algorithms per se are not patentable. I'm guessing that Unisys said something like, "We patent the use of LZW for all computer images," and the patent office bought it. I don't want to waste the time researching it because I don't intend to pay lawyers to fight it. Instead, I'll just gripe to you about it and maybe you'll refuse to do business with the evil Unisys empire.
Can you tell I don't like Unisys? We planned to use GIF in some of our software, but Unisys lawyers said that we had to have each one of our end users negotiate an agreement with Unisys before we allow them to read or write GIF files. I asked why they required it of us and nobody else, and they said that other people had either paid them more than half a million dollars, or they were in violation of their patent. This was for internet software. Linda Timms and Nathan Cass, both Unisys Lawyers, kindly suggested that we remove the internet capability from our software to make the GIF licensing easier. Sure thing.
There are literally hundreds of software packages that read and write GIF files for the internet. I am pretty sure that more than half of them are not paying Unisys a dime. In fact, Microsoft's Visual Basic reads GIF files. If you dig deep into their online documentation, you might eventually find the hidden paragraph where Microsoft says Unisys "may" have a valid patent on GIF file reading, so if you intend to distribute your software commercially you should contact Unisys for licensing arrangements.
Everyone who writes software that reads and writes GIF files is supposed to pay royalties to Unisys. This patent is very selectively enforced. Unisys generally lets a small company get bigger and then sues them when they have enough money to mess with. There are supposedly non-infringing methods of reading and writing this format available, but I think Unisys is happy to sue anybody who uses GIF regardless of the algorithms used. I recommend that everybody use the PNG file format for web graphics instead of GIF. The only disadvantage is that Netscape doesn't recognize the PNG transparent color yet, but they will soon.
If you go to the website www.gm.com, you expect to find General Motors, right? www.ibm.com should take you to IBM's web site. There are companies who realize this, and register thousands of domain names that are similar to familiar company names. They're called cyber-squatters. Then when the familiar company wants a web site, they have to pay the cyber-squatters a whole lot of money to use the domain name. This is kind of an abuse of the web domain registration process. Congress realized this, which is pretty surprising, and the House passed a law against it Tuesday. But Clinton, who gets LOTS of money from trial lawyers, may veto it. He says, "The best place to resolve domain-name disputes is in the courtroom." Right.
Speaking of IBM, in 1931 they sold the first IBM machine that did multiplication. Yesterday IBM dedicated a supercomputer a Lawrence Livermore Labs that does 4 teraflops. One flop is one floating point operation per second. A floating point operation is something like the multiplication or cosine of a real number. A megaflop is a million floating point operations per second. This computer does 4 million megaflops -- 4 teraflops. That's FAST. It does it through parallel computing with 5800 processors. The computer was installed about 2 years ago. It took them this long it get it up and running this fast, believe it or not. Intel currently has built the only computer faster than this, but next June IBM plans to install a computer at Livermore Labs more than twice this fast -- 10 teraflops. The IBM history is pretty interesting. Check out
This junk mail may be protected under United States Patent No. 4,558,302.
Or it may not.
Unauthorized copying of this wonderful pseudo-literary word collection just doesn't matter.