Dr. Jerry Pournelle

Email Me

Why not subscribe now?

Chaos Manor Subscribe Now

Useful Link(s)...


Hosting by

Powered by Apache

Computing At Chaos Manor:
May 15, 2007

The User's Column, May 2007
Column 322, part 3
Jerry Pournelle jerryp@jerrypournelle.com
Copyright 2007 Jerry E. Pournelle, Ph.D.

Continued from last week...

The column schedule for the next few weeks is going to be hectic at best. WinHEC opens Tuesday of this week. Niven and I are within 3,000 words of the end of INFERNO 2, and that book must be finished. Then at week's end Niven and I are off on a two-week trip to a Homeland Security Conference, then to be Guests of Honor at Balticon Balticon over the Memorial Day Weekend, then back to Georgetown in DC for conferences with our editor, and possibly a book signing.

I'll have a computer with me, of course, and I'll try to keep up, but things can get hectic. In particular, next week's column is likely to be delayed due to my GOH duties at Balticon. Reverberations from that may last for part of June. This is a big trip.

The good news is that Peter Glaskowsky has edited Two Steps Farther Out, the sequel to my A Step Farther Out. He has left notes where my attention is needed, and I have some work to do but I hope that within the week it will be available to subscribers. Like A Step Farther Out, this is a book made up of columns and essays written some years ago about the state of science and technology in the United States. Alas, many of the political issues that affect scientific problems covered in those essays remain with us. We didn't make the right choices back then. In some cases we made exactly the wrong ones. There is some "I told you so" in the book, but not much; mostly I just want us to get back on the right track.

Despite the work's age, the essays are pretty current. I am not revising the book, but I am inserting a few annotations to bring things up to date. It has needed surprisingly few. When I get it finished it will be posted in the subscriber area.

Efforts are also under way to get A Step Farther Out (Originally published by Ace Books in the 1980's and in print for over twenty years) into electronic form to be posted in the subscriber area. It too will have annotations rather than revisions. We'll be adding other material for subscribers in the next few weeks.

The Empire Strikes Back

It has long been known that Microsoft President and CEO Steve Ballmer has a thorough dislike for Linux, GNU, GNU/Linux, Open Source Software, and all their kin. (link) Some of this was generated by the ill starred "campaign" by Linux users to turn in Windows for a refund (link), which was seen by Ballmer and his companions as deliberately twisting the lion's tail. Most of us have expected Microsoft to strike back at Linux enthusiasts; the question was, how might they do that?

The shoe dropped last week: Microsoft charges that Linux violates no fewer than 235 Microsoft software patents, and Microsoft expects someone to pay for these patent infringements. (link).

It's not entirely clear just who is supposed to pay for what; in fact, it's pretty clear that Microsoft hasn't figured out that detail. Linux software doesn't have a single author, and while companies can sell Linux distributions and charge for support, the software itself is free, and can be freely copied. This makes it hard to find targets.

The two main characters in the GNU/Linux drama are GNU founder Richard M. Stallman, and Linux kernel author Linus Torvalds, and while both of them have been vital in making Linux successful, neither has gotten rich by doing it: they do not between them have enough money to pay the licensing fees Microsoft will be due assuming the validity of the patents.

A number of big companies that use Linux certainly have enough money, but they are also Microsoft's largest customers; and few companies have got rich by suing their customers. Then there's the threat of the renewal of the monopoly prosecutions. The current Justice Department isn't inclined to do that, but that can change. If Microsoft alienates big corporate customers, who in revenge turn to Linux - about the only possible competitor just now - and then Microsoft uses its patents to preclude their using Linux, the monopoly charges are certain to be dusted off.

Up to now, Microsoft has been satisfied with using its patents to obtain cross-licensing agreements. One such was with Novell. There are also said to be agreements with Sony, Sun, Siemens, etc. Note that the actual agreements are not really formal cross-licensing contracts; instead, each agrees not to sue the other or the other's customers for patent infringements. There is a reason for this peculiar wording and we'll get to it below; the practical effect, though, is cross-licensing.

As to what's at stake: Microsoft Lawyer Horacio Gutierrez says that the Linux kernel violates 42 Microsoft patents. The Linux graphical user interfaces infringe another 65, and the Open Office suite of programs infringes 45 more. Microsoft hasn't yet gone public to announce specifics, but it's pretty clear they know what they're doing; this isn't anything like as vague as the SCO infringement claims.

I note that Linux enthusiasts - most of whom tend to be hostile to Microsoft - say that this is precisely like the SCO case. One even says that Microsoft has instituted these direct attacks on Linux because Microsoft could no longer use SCO as a front for harassing Linux. I don't accept that. I suspect that Microsoft's policy of paying fees to SCO stemmed from more complex motives than that, and no one really expected the last-ditch SCO suit to succeed. SCO was headed for the skids; despite Microsoft's enemies wishes, that's just not true of the Redmond giant. Microsoft has cash, smart lawyers, and, when needed, dogged determination. Linux supporters would do well to remember that.

Incidentally, one of my advisors thinks it silly that I mention the "turn in your Windows" stunts of a few years ago as a factor in the Microsoft decisions. I don't think it silly at all: I've known both Gates and Ballmer for many years, and with them sometimes it's personal. Machiavelli warns us never to do an enemy a small injury.

The Courts

Will these claims stand up in court? Microsoft's lawyers certainly seem to think so. Suppose there is an 0.5 % chance that each would be upheld. With 235 shots, the probability isn't small that at least one will be a winner; and of course many of those have a lot better chance than that. You may be sure that Microsoft will have a good story, and will have the resources to present that story well.

Then we must ask, who can stand up to Microsoft with all its resources? Just who will finance the opposition? Microsoft seems to be offering fairly generous terms, and already has a number of takers. Few big companies value principle enough to engage in a big fight when it's easy enough to buy your way out. The obvious opponent is IBM, which does get a lot of value out of Linux; but IBM hasn't much to gain from a dog fight with Microsoft. It does have a lot to gain from a quiet settlement that allows everyone to get on with what they do best. I would have to see a lot more detailed analysis of what IBM has to gain, financially, from getting into this fight before I would conclude Big Blue will enter the arena.

Open Source

The whole issue is complicated by Richard M. Stallman's ingenuity in drafting the free software licensing agreements. The licensing terms for GNU/Linux forbid paying anyone patent fees for using the software. Now third party agreements can't deprive Microsoft of its due - assuming the patents are valid and have been infringed - but they can prevent anyone who pays Microsoft a patent fee from using GNU/Linux; or at least GNU/Linux enthusiasts fervently believe so.

Note I said GNU/Linux, but in fact much of Stallman's language has been incorporated into other Linux agreements as well. Stallman, understandably, insists that there is no "Linux", only "GNU/Linux" and refuses to talk to people who believe differently. I've known Richard since he was a student at MIT, and I am hardly astonished. The point is nearly moot: the most usable forms of Linux definitely depend on work coming from GNU. For those confused by all this, do some Google work on the history of GNU and Linux; it's too long to repeat here.

Software Patents at Stake

There are more wild cards. The US Supreme Court has not yet given a definitive and final decision on the whole notion of software patents. The concept of patents on software was quite controversial in the early days of microcomputers, and most software was copyrighted, not patented, until about 1990. Many computer writers and legal theorists adamantly opposed the whole idea of software patents.

Detractors of software patents claim that software is more akin to a mathematical algorithm than an invention, and the courts have ruled that algorithms aren't patentable. Note that patents are far more general than copyrights: patents protect methodologies, step by step instructions on how to accomplish a given result, and not just the particular implementation at hand. An analogy might be if I could obtain a patent on the Odyssey plot, and demand license fees from anyone who tried to write a novel about the adventures of a man coming home from the war; this as opposed to my copyrighting a novel with that plot. My copyright would protect only the work, not the idea behind the work. A patent would protect both.

The recent SCOTUS decision on patents has probably scotched the notion of the plot patent. It said nothing about software patents in general. The Supreme Court could, reasonably, hold that software patents are indeed akin to algorithms, and thus cannot be patented at all: software can be copyrighted, but software procedures cannot be. You may expect a very great deal of very expensive litigation before this issue is resolved.

So: we have Microsoft claiming that GNU/Linux and the entire free software movement infringe on Microsoft patents. We have GNU/Linux claiming that no one who uses their software can pay patent or licensing fees to anyone in general and Microsoft in particular. We have some theorists who see this as a marvelous opportunity to challenge the entire notion of software patents.

The one certainty here is that a number of lawyers will have scads of billable hours. For the rest of us, it's going to be interesting times. I have a number of comments from many sources; we'll continue this next week.

Outlook Ups and Downs

Last week I mentioned that piggy old Outlook was more porcine than ever, and that so far as I know I hadn't done anything to cause the problem. That turns out to be correct on both counts.

One of Microsoft's automatic updates caused the problem. You can find the details here. After a few days the problem fixed itself, which is to say, Microsoft apparently shipped out another software update that undid the evil effects of the last one.

Meanwhile, I did make one adjustment on my system: I set Outlook to use just one processor in my dual processor system. To do that in both XP and Vista, open Task Manager, Processes, and find the Outlook process. Right click on that, select properties, and open set affinity. You may now assign access to either or both CPU's; the default is both. In my case I told Outlook it couldn't have any of the second processor; it had to make due with Processor One.

This had no noticeable effect on Outlook either while it was in the ultra-piggy mode or after the update fixed the problem, but it did free up the second processor for Word, FrontPage, and other applications. Prior to restricting Outlook, I could be typing in Word when the system would effectively halt for from one to ten seconds while Outlook used up all the resources. Now Outlook has to leave something for other programs.

I have left Outlook restricted to one processor even though the glitch has been fixed.

One does wonder: here we get dual processor machines in the hopes that we can do multi-tasking. Microsoft, though, sees multiple CPU's as an opportunity for its minions to grab even more resources and tie them up. Of course that does no good: when Outlook glitches it's not CPU bound, it's hung up on an I/O operation, with the CPU merely going about in circles until the I/O operation is complete - but that doesn't stop it from taking over all the CPU cycles it can grab. I would not, myself, think that is the best possible way to write a program.

For all my complaints, Outlook does work, and sometimes it works well. Coupled with Microsoft Desktop Search, which knows how to build indices of those enormous unparsable .pst files, Outlook can be really cool. The other day I needed to find an email sent to by Roland Dobbins, an old friend and frequent advisor. Since Roland sends me up to a dozen emails a day, and the one I wanted came several weeks ago, finding it could have been a problem; but it was a commentary on Inferno, so I typed Roland + Inferno into the Microsoft Desktop Search window; and instantly there appeared the exact email I was looking for. It took less than a second.

Outlook sometimes infuriates me still, but setting it so that it can't interrupt everyone else doesn't slow Outlook down much, and it's sure good for my blood pressure.

Lenovo X60 TabletPC with Vista Pro

The other day I had to drive someone out to Kaiser for a medical appointment. It looked as if I'd be spending a couple of hours in a waiting room. Larry Niven and I are on the last two chapters of Inferno 2, and we had reached a critical place. I really wanted to work on that book, so I loaded it into the Lenovo X60 TabletPC, on the theory that I would not find a table or desk to work at.

We've been working in Word 2003 on XP systems. Leonardo, the Lenovo TabletPC, has Vista Professional and Office 2007. That proved to be no problem. Word 2007 opens and saves Word 2003 documents without incident. It's a bit confusing to go back and forth from Word 2003 to Word 2007 - the tools, menus, and general layout are entirely different - but it's not really difficult.

As I had anticipated, I had an hour or so to wait in the waiting room. Uncomfortable chairs, no table, not really a place to put a laptop on my lap and try to pound keys. Instead I rotated the screen so that Leonardo was a tablet, no keyboard visible, opened Inferno, and began to work on the book.

First I read through a couple of chapters I did last week. Naturally I wanted to fiddle with the text, so I tried the tablet features. Voila! Everything worked perfectly. If I write carefully, Vista Tablet recognizes my handwriting amazingly well. I can insert words, change words, and it's almost like doing it in pen and ink on a printed manuscript. Then I came to the end of the text. I thought of a sentence, and began writing it. That worked. I did another.

By the time we were ready to leave I had written about 500 words, and edited two chapters.

I still prefer the HP TabletPC form factor, and I particularly like the HP's jog wheel; but then I've been carrying that TabletPC for years, and I am very accustomed to it. The Lenovo X60 is a bit heavier, and somehow doesn't feel as graceful as the HP, but that may be a matter of getting used to it.

The Lenovo X60 is far faster than the HP TabletPC - with a Core 2 Duo it ought to be - and the screen is both brighter and more readable. The battery indicator told me I had 5 hours battery life, and as I used the system, the estimate lessened in real time: that is, after two hours work it said I had 3 hours left.

The screen is smooth and easy to write on. I have both the built-in Lenovo stylus and a Wacom Cross stylus which I prefer. Both work just fine.

All told, it was a delightful experience. I would probably have got more work done if I could sit at a table and type, but the point is that I did a respectable amount of work sitting in a chair using the pen for input. I'll be carrying that Lenovo X60 on my trip. Full report when I get back.