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:
February 12, 2009

The User's Column, February, 2009
Column 343
Jerry Pournelle jerryp@jerrypournelle.com
Copyright 2009 Jerry E. Pournelle, Ph.D.

Most of the economic news is bad, but there one thing remains: technology advances continue. Moore's Law prevails. We are very much still on the rising section of the "S" curve of technology. (For more on S Curves, see The Strategy of Technology by Possony and Pournelle.) For more on the future of technology in 2009, find L. Gordon Crovitz's report on the recent Technology, Entertainment, and Design conference.

I used to go to the TED conferences back in the early days of the Computer Revolution. Most of the goshwow computer booster conferences have died away, but TED remains, and there's still some excitement. It's pretty clear that we have the technology to implement at least one more revolution comparable to the microcomputer and World Wide Web revolutions. All we have to do is get on with it. There's no real agreement on just what that revolution will be, but there are candidates.

One of the candidates for the next revolution discussed at the TED conference was a presentation by Tim Berners-Lee, generally acknowledged as the inventor of the World Wide Web. He's calling for a data revolution: a way of putting the enormous floods of data now out on the Internet into compatible formats so that relationships among wildly different kinds of data can be found.

This would allow new correlations to be found, and those may lead to scientific breakthroughs by generating new falsifiable hypotheses to be tested. Note that discovery of a correlation isn't itself a breakthrough, just as finding improbable events isn't necessarily a new discovery. Think about J. B. Rhine's work at Duke, where undergraduates used the Rhine cards to search for telepathy and thought they had proved the existence of Extra Sensory Perception - ESP - beyond the shadow of a doubt. They certainly did find results that were highly improbable, some results with probabilities as low as 1 in 100,000; pretty impressive until you realize that there were over 100,000 replications of the experiment. Given enough replications, the improbable isn't just probable, it's inevitable.

Rhine's stellar performers would of course go on to be no more successful at telepathy than anyone else: ESP believers are still searching for why they lost their psychic abilities, but the more logical explanation is that they never had them, and the laws of probability operated precisely as one might expect. Note, though, that I say this as a caution: I am pretty sure that getting more data into compatible formats will result in a number of scientific advances. All we need do is apply the scientific method: data mining generates hypotheses; sharpen those into falsifiable theories, and then test the theory. It's not as dramatic as examining entrails, but it works better.

After all, the Internet Revolution has already produced Rev 1, which is Google, and the fallout from that has tremendous impact - not only on those who use Google, but on technology developments, some of which is still very proprietary. As an example, BYTE once presented its Best Technology Award to a company that had developed techniques for clustering large groups of servers. Google has gone way past that now, although you aren't likely to get any Google engineers to tell you much about how they do it.

Another benefit of the new technology is computational plenty. When I was an undergraduate I was invited to the University of Illinois to see and play with the ILLIAC, then the most powerful computer in the world. It was housed in what had been a basketball gymnasium and consisted of row after row of long files of racks of vacuum tube units. Three undergraduates had the task of patrolling down those aisles with shopping carts full of vacuum tubes: they'd spot burned out tubes and replace them. Because tubes could burn out during a computation, ILLIAC routinely did each calculation three times and took a majority vote on which answer was correct.

Today, when one of the tens of thousands of Google servers gets sick, it sends a signal. "I'm sick," it says. "Fine, shut up and die," it's told, and it takes itself off line, sitting there until someone can get around to replacing it. That in itself is a revolution in maintenance. Enough redundancy begets a qualitative change.

I don't know what the next breakthrough revolution will be. My friend Phil Tharp points out that tools that allow one person to control another's computer across the web enable us to do things we never could do before, and may well be a step to a revolution beyond the distributed computing revolution our micro computers sparked. I think he has a point, although neither of us, so far, has been able to see quite where that may lead, other than the criminal zombification of hundreds of thousands of machines. There has got to be a more useful application.

Computer technology advances. We're on the steep part of the S curve, which means that everything gets better at breakneck speeds. That applies to robots as well as thinking machines. Computers already are used to generate improvements in the next generation of computers. Roy Kurzweil believes we're approaching a point when computers will be turned loose to do that for themselves, bringing on not merely a revolution but the Singularity predicted by writers like Vernor Vinge. A singularity is a revolution whose outcome cannot be predicted and may not even be imagined. One doesn't have to believe that's coming to believe there are going to be profound changes in our society as a result of all this computing power.

The history of the world can be summed up by the observation that the usual course of civilization is to devote larger and larger parts of its productivity to the construction of structure, until the structure becomes so rigid as to stifle growth and change. Sometimes, though, productivity leaps ahead faster than the regulators can move. The Discovery of the New World, the First and Second Industrial Revolutions, the computer revolution, development of the World Wide Web, all were such events. Technology and productivity leaped far ahead of regulators, who are now catching up; but technology will probably move faster.

Peter Glaskowsky adds:

Artificial cognition is already here in a limited way, and is starting to shape up into a detectable curve: cars can maintain a safe following distance and park themselves; digital cameras can take pictures when all the subjects in the frame are smiling and their eyes are open.

I see no limit to how far this will go, and I would estimate that artificial cognition will influence a trillion dollars in spending over the next ten or fifteen years. Close enough to a revolution?

I use that term rather than "artificial intelligence" because I don't see true AI as a problem we're going to solve any time soon, but artificial cognition is just a function of sensors, computers, and programmers being told what the sensors and computers should look for.


Artificial Cognition is very likely to be part of the Next Big Thing, and should be easier to come by than the full blown Artificial Intelligence that we've pursued for decades. AI itself is a bit like the will-o-wisp: it recedes as you approach it, and you never quite catch it.

Of course, we have had computer based expert systems for years. I recall one that looked at railway wheels to determine when it was time to replace them. Others are more complex. At one time there was a flurry of activity in medical diagnosis using small computers, and a program, Tieresias, was actually in use for a while. It recorded a lot of data and then applied rules to suggest diagnoses. That was on Z-80 based systems, and it was slow and limited; we have come a long way since then, and I expect there are a number of people looking into ways to automate medical history inputs, then apply rules to suggest diagnoses. We certainly have the technology to build large expert systems making use of what Glaskowsky calls artificial cognition.

There are other ways to produce revolutions. Reliable and economic access to space would generate another productivity revolution, but that may not be necessary: that is, it may be the consequence of events for which we already have the technology we need. I don't know what the next revolution will be, but I'm pretty sure it's coming.

Bill Gates at TED

One of the talks at TED was given by Bill Gates. It's very much worth your while to watch it. Gates looks at two world problems, malaria, and how to improve teaching. His views on education are different from Charles Murray's Real Education: Four Simple Truths. Gates seems convinced that a 4 year college education is important to everyone, and that everyone can benefit from that. I am not at all convinced of that.

On the other hand, Gates has data on teacher quality; the best teachers can do wonders, not just with bright students but with all of them. If the majority of our teachers were as good as the top 20%, there would be an enormous gain in educational effectiveness across the board from dull normal pupils up.

Gates also has data on what produces teacher quality, including the value of a Master's degree for raising teacher performance (essentially zero) and the value of seniority (not much better); those are, of course, the major factors in teacher pay raises. Performance is not counted at all. He believes that as the data revolution continues, people will catch wise to what's going on, and we will find ways to reward good teachers for better performance, thus retaining the best (many of whom now drop out of the system). He also believes we can teach others how to perform as well as many of the best teachers. The resulting education revolution would certainly change the nation.


The Kindle 2 is out, and I'll get one. I'm not entirely sure why; probably a return of my old fascination with new technology, and partly because I do lots of things so you don't have to.

Mind you, I don't consider getting any Kindle a silly thing. I have been using my Kindle 1 for a year now, and I wouldn't be without it. I use it for at least half of my reading. I haven't yet tried it for newspapers because I only read newspapers at the breakfast table, and the home delivery system works well here, but I do admit that on Tuesday nights when it's time to take out the trash for the week, I could do without staggering out with a week's load of newspapers for the blue recycle can. Of course most trees used to make newsprint are planted for that purpose, and if no one is buying trees for paper pulp then no one will plant trees on those tree farms; they'll find some other crop for that land. Perhaps they'll turn it into cattle range, and the resulting methane will help warm things up; cold as it is here just now, I could use some warming.

I haven't seen Kindle 2, and my initial reaction was that it wasn't enough better than Kindle 1, particularly since the price remains $359; and I don't really advise current Kindle users to rush out and order the new one, since the price is very likely to fall within a year, and the old Kindle works pretty well for me. The new one does use the 3G network, which will be faster, and has a somewhat larger screen. The controls are perhaps a bit more instantly user friendly (although it only took me an hour or so to get used to the Kindle 1 controls: mostly by learning how to pick it up without turning pages).

In any event, if you don't have a Kindle, you might think about getting one. On my last trip, I carried my Kindle on the airplane, and thus had my choice of more than a dozen books I could be reading. For that matter, the radio purchasing service worked fine on the airplane: I probably could have bought any other book Amazon sells for the Kindle while in flight. Moreover, there are many other sources - like Baen - of books that are easily downloaded on a PC and then transferred to the Kindle. Of course, I can't use the Kindle for editing or writing; I need a TabletPC for that, or I can carry the MacBook Air in the hopes that the chap in front of me won't lean his seat all the way back; but in fact I don't do much work on airplanes now. I'm content to sit back and read books, and the Kindle is excellent for that. It's not back-lit, but neither are books, and in fact the Kindle is more readable in airplane seat lighting than most books. If I travelled as often as I used to, I'd really, really want a Kindle.

Kindle 2 is somewhat better than Kindle 1. The screen is larger with more gray scale shades (which should make it a lot easier to read newspapers on it). It's said to be somewhat easier to make notes with, and the battery life is supposed to be longer. Apparently it does not have an SD memory card slot, for reasons I don't understand; that would be a disadvantage. Otherwise I don't know the comparative advantages and disadvantages of Kindle 1 and Kindle 2, but for the moment I don't think the advantages are so much greater that you ought to rush out and replace your Kindle 1 just yet. Stay tuned on that.

I do know that since I got my Kindle 1 I haven't bought a book in an airport, and I have bought very few books in paper that were available on the Kindle. I have posted pdf documents to the Kindle, - I email them to myself at my Kindle address, and they appear on my Kindle; Amazon charges me a dime for doing that - and put over mss. sent to me for evaluation. I can't edit or proof read on the Kindle, but it's sure easier to read books on it than to carry hundreds of pages of paper. The pdf document transfer isn't perfect, but I put Strategy of Technology on my Kindle and I can read it; the notes are handled badly, and it's difficult to skip ahead and find things, but it's readable. Of course few books are as complex as Strategy of Technology to begin with, and the pdf was done long ago.

Kindle isn't without competition. Many prefer the Sony reader and there are others out there that people like. None have, in my experience, the simple ease of use including ease of purchasing new books that the Kindle has; Amazon put a lot of thought into that and did it well.

Whatever the fate of Kindle 1 and 2, Amazon has shown the path that leads to a great reduction in the importance of cheap words on paper publishing. EBooks have two effects on books: they're a lot cheaper than hardbound books (Kindle books are usually about $9.95 for a $30 hardbound), and you can carry a great number of them in place of paperbacks. It is estimated that the New York Times could give each subscriber a Kindle and save money on printing and distribution of the newspaper. Layout design hasn't managed to make it easier to read the Times on Kindle than on paper, but that could change.

We may not have seen the exact picture of the future, but it's pretty clear that Amazon and Kindle are leading us somewhere that we are bound to go, and the effects on the publishing and entertainment industries will be profound.

HP MediaSmart Server

I recently got rid of my Windows 2000 Server Active Directory Network, and converted the Chaos Manor establishment network to a Workgroup. That has worked well, and I don't really regret it, although the various security settings of Vista are maddening: one of my machines can pull data from any machine, but nothing can push to it. I am sure there's a firewall or security setting I can't find that's doing that, but I don't know what it is. One Vista machine can log on to any other Vista machine automatically, but another has to have that logon done manually by providing user name and password any time after the machine has been shut down. Once again I have no idea what settings are different on the two systems; I suppose I'll have to make time to do a systematic search, but so far I just live with it, since once the logon has been done, everything else works all right until the next restart.

Once the conversion from Active Directory was done, I was ready to try Windows Home Server. I'd seen that in action at both the Professional Developers Conference and WinHEC, Microsoft's back to back technology conferences last fall in the Los Angeles Convention Center. I went to those conferences in a bit of a fog: my brain was still sunburned from 50,000 rad of hard X-ray treatments for an inoperable lump in my head, so I didn't get as much out of the conferences as I usually do; but they were well worth my going to, and I did go to sessions on Windows Home Server.

Windows Home Server is rock solid, based on Windows 2003 Server. It provides three services: automatic backup, restoration, and health monitoring of one of more machines; document and file sharing; and remote access capability from anywhere you have Internet access. Shared files can include photos, audio, and video. Backup is complete, and incremental, and can cover as many machines as you like. I've scheduled daily backups at 0400, and it all goes smoothly. For a full review of Windows Home Server with considerable attention to details, see this link.

You can get Windows Home Server from Microsoft - there's a 120 day free trial offer - and install it on a "box of drives" system you build or buy; or you can get a third party system with the Windows Home Server installed on the box along with installation software to install on each machine Home Server will serve. I chose the HP MediaSmart Server, in part because the HP system includes wizards that make installation very simple - which is just as well, because I haven't taken the time to learn Vista as well as I should.

The HP MediaSmart Server is quite handsome. It's about the size of two shoeboxes. There's no screen or keyboard: the server sits under a desk or in a closet, and is controlled by the machines it serves. HP sells MediaSmart Server with up to 4 terabytes of storage: the box will hold four drives. In my case I have several Western Digital Terabyte drives, so I only bought the basic unit with one drive. Installation of new drives is simple, and can actually be done after the system is set up; you can even hot swap drives on the HP server.

Installation was extremely simple: just plug in the server to power and the Ethernet, then insert the disk into the machine you'll use to control the server. A wizard comes up, and you just follow instructions.

A few cautions. If your current system has a short and simple minded password, you won't be able to use that for the Home Server, which is going to insist on at least 6 characters, and really wants a strong password. Moreover, if you intend to allow access to the Windows Home Server from the Internet, you'll want a very strong password for that access; fortunately you can use a different user name and password for that operation.

When you're finished with the system, it will appear useless, or it did for me. Once I finished with the wizard installation I wasn't able to do anything with the server - I named him Jeeves - at all. I couldn't copy files to it, and I couldn't figure out what was wrong. In frustration I took a short walk; when I came back all was well. Apparently it takes Vista a while to digest the Home Server. Once it has done that, it works as expected.

I find Jeeves very simple to use. We don't do a lot of media sharing at Chaos Manor, but it would be simple enough to do so. Unlike a lot of new stuff, the HP MediaSmart Server Just Works. I haven't done a lot with it other than stuff a new drive in while it was running - it takes care of that automatically and without fuss - but it sure solves the automatic backup problem.

After the initial bafflement while Vista got used to having Jeeves on the system - less than half an hour - nothing else of much interest happened, and thus I don't have a lot more to say. The bottom line for me is that if you need a home server - and most of you really do - this one works, and the price is good. Recommended.

FileMaker Pro 10

If you're looking for a data base, and most of us are even if we don't know it, use FileMaker. I'll do a full review in another column, but I have no hesitation in recommending FileMaker; I've been using earlier versions for years. Unlike the company's simpler Bento product, FileMaker works on both Mac and Windows systems, and thus is especially useful for people like me who use both. If all you need is a good data base for a single Mac, and you're not going to try data mining in collaboration or across platforms, I suppose Bento would do, and it's certainly more Mac-like; but the full-blown FileMaker Pro doesn't cost that much more, and while I don't collaborate with others on building and mining data bases, I do collaborate with myself - that is, I use more than one machine, and I'm moving some of my operations to the Mac OS. That latter has been stalled a bit. One reason is that my most powerful machine is an Intel Core 2 Quad Extreme that handles multiple tasks like lightning - it's the only machine I've ever had that Outlook 2007 can't make stutter when it's downloading a big message and looking at it with a complex of rules. If I had a Mac of similar power I'd probably change over all the way. As I get older, Vista's arcane security settings - see above for some examples - have become more and more irritating. I understand that Microsoft has the problem of satisfying old customers while trying to make improvements, and by and large they've done well at it. And, of course, Windows 7 seems to be what Vista ought to have been, so we'll see.

On the other hand, I have a lot of incentives to do some video and audio podcasting, and from everything I can see, Mac software is much better for doing media recording and editing - better in the sense of ease of use, of course. I'm not qualified to comment on final quality of products, but then I don't have to be. Good enough is plenty for what I need to do.

In any event, I am pretty sure that both PC and Mac systems are in my future, which made it easy to decide on using FileMaker 10 rather than Bento as the Chaos Manor data base. I'll have more to say on FileMaker tricks and uses another time. Recommended.

Winding Down

The book of the month is Charles Murray, Real Education: Four Simple Truths For Bringing America's Schools Back To Reality (2008 Crown Forum). Murray and Gates ought to be taken together. Gates hopes to improve the schools to the point where a much larger fraction of the population can profit from a college education; Murray shows why this is both unlikely and undesirable. Both agree that we need better teachers. Neither is likely to be popular with teachers' unions. I consider the breakdown of our education system the greatest existing long term threat to the United States. It's much more serious than anything going on in the Middle East, and just now there's no remedy anywhere in sight.

I'm glad to see people like Bill Gates interested in the problem, although I think he's on the wrong track. Murray is one of the few sociologists whose work I trust and admire. Between Gates and Murray we may find the means to change the education system so that it serves both those who should be in colleges and universities, and those who should not be.

The computer book of the month is Rachel Andrew and Kevin Yank, Everything You Know About CSS Is Wrong (2008 Sitepoint). I don't use Cascading Style Sheets (CSS). My web site ( http://www.jerrypournelle.com ) is built with FrontPage, and even though that product is no longer supported by Microsoft, it remains Good Enough for what I do; but then I have a very unfancy web site, based on content, not on whizzbangs. It has been a long time since I won any awards for site design, but my users don't seem to mind.

Many web designers and webmasters do use CSS. In the words of the book, "It gives web designers a language in which to describe a consistent visual treatment that can be applied to a single page, an entire site, or even a whole bunch of sites." If you use CSS or are considering it, you will need this book, which tells the story of how CSS was founded in the days of the browser wars, and how Microsoft's failure to keep IE up to date with tools like CSS allowed Opera and Firefox to stay alive after Internet Explorer pretty well killed Netscape. It also deals with CSS's limitations and failures, and the various tricks, techniques, and just plain kludges that web designers use to get CSS to do things it wasn't really designed to do.

If, like me, all you do on the web is write content and put up a few pictures, you don't need this book; but if you do web designs, or especially if you're thinking about getting into that business, you very badly need this book, since CSS is one of the tools you should be familiar with. I'm keeping my review copy in case I ever get the crazy urge to do some web design myself.