November 29, 2003:

Slashdot recently aggregated an article from Red Herring on the economics of file-sharing that may stand as one of the most articulate (if tricky to follow) arguments against the current business model. It analyzes the current music situation as a "double moral hazard" (this is economics jargon) and suggests various ways out of the trap, none of which strike me as especially likely to be adopted by anyone on either side.

In more file sharing news, several indie bands are infuriated at being listed as members of the RIAA when they're not. (See this short item from TechDirt.) They're running afoul of something called RIAA Radar, which is basically a blacklist of artists who belong to the RIAA. Is this notice of false membership fraud, trademark infringement, or defamation of character? An upcoming lawsuit may attempt to find out.

The future mode of file sharing will almost certainly be: Gather globally. Share locally. I continue to be astonished at the sheer quantity of material of all kinds posted on the binaries newgroups on Usenet. If you wait long enough, almost anything you'd ever want will appear—and with cavernous 200 GB hard drives to store things, file sharers will gather indiscriminately, probably with robot newsreaders, and assume that they'll never listen to most of the music files they grab. At LAN parties and other local gatherings (including P2P meetings held among Wi-Fi equipped laptops) they will use utilities like Lanster to browse one another's hoards and pick off what they want that they don't already have.

I had an idea the other day, doomed though it might be: Suppose a new law either allowed or required the creation of an online licensing agency for music, something like a next-generation ASCAP. People could buy an annual subscription to the agency, and in return receive a small application (I call it the Conscience) that they install on their main systems. The Conscience would scour their hard drives for MP3s, and report statistics on what tracks were present (anonymously) to a central server. The track distribution data would be used to distribute licensing proceeds among the artists. The law would immunize people who subscribe from copyright infringement liability, period. No quibbles. Track distribution data would also be sold to whoever wanted it to support the administrative overhead of the system. (Radio stations and record companies would pay dearly to have such a razor-sharp picture of what sort of music people favor.)

A system like this would have a number of effects:

  • The artists themselves would receive more money for their work than they do now. (Anything is more than nothing!)
  • The album as a format would die. It has been horribly abused for so many years simply to keep CD prices up that this is inevitable. I won't miss it.
  • Big Honking Record Companies would shrink radically, though they'd never entirely vanish. I won't miss them either. Savvy bands would become their own publishers and eliminate the middlemen.
  • Record stores would mostly vanish. This is sad, but in truth, record stores are ill-equipped to deal with a music industry as large and diverse as ours is now. They don't have the shelf space or the capitalization to sell half a million different tracks.
  • Obscure bands would have as good a shot at making money as famous bands. Promotion would fall more to the bands than to record companies, but at least obscure bands would not be subsidizing promotion of popular bands, as happens now.

This system isn't entirely original with me; it's called "compulsory licensing" and people suggest it regularly as a solution to the current viciousness raging over music copyright. The Conscience is the twist I added. Would people pay for a subscription to something like this? I think so. $75/year sounds about right to me, and I'd pay it in a second. Would they use the Conscience? Perhaps...if they thought it would mean more money for their favorite bands, especially if those bands are not Top 10 fodder, or if running the Conscience gave them discounts on concert tickets or other perks.

Freeloading would be an issue, and it may come down to tricky propositions like levying a $6/month licensing fee on broadband Internet access ("compulsory" cuts both ways) which would be eliminated for those systems on which the Conscience was running and found no music tracks. I'm not saying this is the best system we can come up with, but DRM and pervasive lawsuits aren't working either.
November 28, 2003:

A few odd lots on the day that reluctant retailers call Black Friday:

  • Michael Covington informed me that his University of Georgia colleague Hugh Kenner died the other day, of heart problems, at age 80. Readers of PC Techniques may recall that Hugh was my book reviewer for awhile, after having done a stint as Byte's book reviewer in the midlate 1980s. His degree was in English and he was best known for his research on James Joyce, but he also studied animated films and computer programming. He was the kind of man for whom the term "polymath" was coined. Requiescat in pace.
  • Per my note in my entry for November 20, 2003: I set up my mailbase on my XP lab machine with POPFile in place for spam filtering, and found that after POPFile had added about 3,000 messages to its database, message bodies started to be lost in PocoMail, particularly those with attachments. The next step is to see if I can import my (immense) mailbase into Pegasus Mail and see if that works any better. Stay tuned.
  • I'm evaluating disk mirroring software capable of automatically mirroring a 256MB USB Flash drive to hard disk, starting with Win2K Professional's built-in mirroring. Got any favorites? Any horror stories? Do share.
  • My Win2K work system (on which I do all my writing) has gotten flaky in a remarkably short period of time (I bought it barely a year ago) probably because I install and uninstall so much software, much of it freeware of uncertain provenance. When I move to the new house I'm going to have two separate machines on a KVM switch: One for daily production work, and an entirely separate machine for screwing around with odd software. This box here is cruisin' for a meltdown, but it will become my eventual Win2K lab machine. The new machine will have as little software installed on it as possible—just what I need for writing, editing, drawing, and programming in Delphi. We'll see if that helps keep it healthy.
  • Not much work expected on the house this weekend, but they're putting the kitchen together this coming week, and it's starting to look a great deal like a house. Good weather this week (highs in the 50s predicted) will facilitate finishing the stucco—but there's still no sign of the rest of my roof tiles!

November 27, 2003:

Thanksgiving Day. I was thinking earlier today about the things I was thankful for, and got around to bravery. Not my own—I'm not sure if bravery is anywhere in my personal arsenal, and I'm less sure I'm anxious for circumstances that would help me find out—but the bravery of others. My father was brave: Even in his early 40s, and with his crooked leg, he waded into a brawl in which several teenage boys were pummelling another boy, and literally threw teenagers larger than he (who stood only 5'6" tall) in all directions to retrieve the victim, and then drove the bruised boy home. I will also credit George Bush for bravery, he who (as the late news just informed me) flew into the heart of the Iraq war zone to have Thanksgiving dinner with the astonished American troops stationed there. Yes, I objected to the war, but that's a separate issue. The guy has guts.

We misuse the word "bravery" a lot in our culture, but I'll offer a definition here: Bravery is the willingness to face danger and death, with full awareness of what might be lost. There was a lot of argument after 9/11 over whether the suicide hijackers were brave or not. Nonsense. They were fanatics, who were certain that God would grant them a harem full of virgins and all the food they could eat (what a sad, sad view of divine salvation!) if they just killed a bunch of Americans. To my mind this is a species of insanity, and madmen cannot be brave. They have no grasp of what's really at stake.

Bravery must be inherent. We don't teach it anymore, but it nonetheless rises regularly from ordinary people who serve in the military, in law enforcement, in fire departments, and in many other places in our society. Bravery may, in fact, be universal. We may all be like the Cowardly Lion, who was brave all the time and just didn't believe it himself until he had to take a stand against the Wicked Witch of the West. Maybe we all just need the proper crisis to bring our bravery to the surface—and what we should, in fact, give thanks for today is that such crises happen as rarely as they do.
November 26, 2003:
Jeff's House Coming?
Great room fireplace
(65K image)
Applying stone facing
(199K image)

Michael Covington responded to the note in my November 24, 2003 entry, which suggested that the rate of technological progress is actually slowing, rather than accelerating exponentially, per the Extropians. Michael agreed, and in his view the rate of progress peaked about 1960.

Now, there's bound to be some quibbling over definitions here. Michael and I are talking about qualitative, rather than merely quantitative, progress. In other words, digital computers were invented roughly 1945—that's a qualitative advance. Nearly everything since then has been a refinement on the same basic Von Neumann architecture. Refinement is, of course, necessary for progress, but unless the new ideas keep coming, I'd call it naive to see us as increasing the rate of progress.

We're certainly riding a wave of some kind—but let's be honest about its shape. As I see it, the Great Century of Invention lasted from roughly 1870 to 1970, with a huge peak during and shortly after WWII. Since 1970 or so we've mostly been improving what already existed. Fundamental concepts like nuclear power and weapons, radar, lasers, jet propulsion, antibiotics, solid state electronics, and even quantum physics, all predate 1970, some by decades. Spread spectrum encoding has been with us since Hedy Lamarr and George Antheil's famous WWII torpedo patent. Television goes back to Philo Farnsworth's work in the 1920s.

A genuine nanoassembler would, of course, be a new breakthrough. I'd even call a nanosorter (one of those nanoscale wheelie things that could pull dissolved molecules of a specified composition out of a water solution) a breakthrough, but my guess is we're a decade or more away from either. Michael suggests that the last truly original technology to go mass-market was the event-driven GUI, and I'm not sure I have any better suggestions than that. I think that the concept of making deliberately simple and cheap computers to do simple things in cheap devices (embedded systems) is also a conceptual breakthrough, dating back onto to 1973 or so, though it didn't happen in a big way until after 1980. Cell architectures for wireless telecomm may be another one, though I'm not precisely sure when it was invented. (It didn't becoming common until the mid-80s.)

One problem is that today's technologies are sometimes invented decades before they become commonplace and begin to affect our daily life, which is after all what the Extropians are actually talking about. Is it really fair to say that the Lamarr/Antheil patent was an advance when it was never actually implemented at the time? And suppose somebody next year figures out how to make the Farnsworth Fusor energetic enough to become an energy source and not simply a laboratory neutron source? Is that an invention of the 1950s or the 21st century?

My touchstone here is to take about two hundred steps back and look at the overall nature of life and how it changes over time. Does 1950 look more like 1900 or 2000? If you think (as I do) that 1950 looks tremendously more like 2000 than 1900, then it's fair to say that technological progress as it has affected the daily life of ordinary people has slowed down radically.

It's an interesting meditation. Feel free to send me your thoughts on it.
November 25, 2003:
Jeff's House Coming?
Assembling kitchen cabinets
(86K image)
Applying stucco
(158K image)

I've been a skeptic on outsourcing from the beginning, and in fact I'm skeptical of any cost-reduction efforts undertaken by companies where the CEO and executive VPs make millions of dollars a year. It's therefore been gratifying to see some recent indications that outsourcing to the Third World has not been an unalloyed blessing, and some companies are bringing jobs back.

TechDirt recently aggregated a report that Dell was bringing back tech support from India for its business customers, ostensibly for their accents, but I have to guess that Dell is seeing some resistance from business customers for exporting jobs. (Not all business customers have this option, and I've heard that many businesses with unexportable fixed costs resent companies that send jobs overseas.) Consumer tech support remains in India. An earlier TechDirt item mentions hidden costs of outsourcing.

In reading recent related items (including some about the latest fad of eliminating fixed work locations entirely for people like programmers and engineers) I'm struck by the inability of management to distinguish between cost savings and productivity increases, which are related but not identical. True, it's much easier to measure costs (like facility overhead) than productivity, and pathologically difficult to measure the effects of things like employee resentment over not knowing where you'll be sitting at work tomorrow, and not being able to have a place to hang a picture of your kids nor keep a few reference books close at hand. I think managers are getting a little too comfortable with their current ability to treat employees as utterly expendable, as in, "Sit where you can, you get no slack for confusion, and if you don't like it there's five people in line who can take your job tomorrow." Word does get around, especially with the Web—do they really think that when times get better, such Neanderthal policies will be entirely forgotten?

Of course, the worst of it is that when such companies eventually tank, there's no way to rub these fools' noses in the fact that their complete lack of understanding of how productive workplaces work has led to their firms' self-destruction.
November 24, 2003:

Frank Glover sent me a pointer to one of Howard Lovy's recent NanoBot entries, concerning the use of carbon nanotubes to construct a "space elevator" (AKA beanstalk) along the lines of Arthur C. Clarke's The Fountains of Paradise. Click to it for the artwork alone, which is stunning.

I'm not sure whether such things are possible, but I've been wrong before—and in this case, I believed enough to work a carbon nanotube beanstalk into The Cunning Blood, though in my case it was one of those "walking beanstalks" attached to an orbiting spacecraft, and used to get people from the ground up to the spacecraft without using electricity. (The McGuffin in The Cunning Blood is a prison planet that has been seeded with minute nanomachines that seek out and destroy electrical conductors, making computing and space travel impossible. Supposedly. But not forever.)

Sad that the biggest hurdle to the creation of something like this could be the threat of terrorism—though I'd wager that the Rifkin-Nader Luddite community would do everything they could to keep it from ever happening.

It's been interesting watching where the progress in nanotech has really been, in a sort of high-precision chemistry rather than individually assembled molecular gimcracks. That's going to be a tough nut indeed, which is why I postulated in my novel that Extropian-style nanocomputers didn't actually happen until almost 2100. We will probably get there, though in this case I don't expect to see it in my lifetime. As I've said elsewhere in Contra, I feel that the rate of technological progress is actually slowing—though keep in mind that I consider "progress" to be something a little more fundamental than simply making faster and cheaper processors that follow a 40-year-old architecture. Cell phones and the Internet are now 20+ years old. (A friend of mine had a cellphone in the early 80s that took up most of a briefcase!) And I'm not entirely sure that our operating systems are any better than the superb VAX VMS, which also goes back 20 years. Windows XP, effective as it is, is mostly a VMS clone in a clown costume. (Does anybody else see the current trend in GUI styles as painfully garish?) These supposed advances are all quantitative improvements in relatively ancient stuff. Where's the really really new stuff? I don't see it.
November 23, 2003:

Reader David Galloway dug deep in his mailbase and retrieved a message I had sent him in January 2000, indicating that I had suspended VDM Diary as of November 1, 1999, "pending resolution of certain issues here." (See my entry for November 21, 2003.) The October entries had been written but never posted, which I recall happening to the December entries, which I evidently never wrote.

I do know that we started having severe budget problems for Visual Developer Magazine in October 1999, and that come December we had decided to pull the plug. (The last issue was March/April 2000.) So I guess that what I have posted here is in fact everything I ever did on VDM Diary. Thanks to David for reminding me just how malleable human memory can be.
November 22, 2003:

John F. Kennedy was assassinated 40 years ago today, when I was in sixth grade. I remember the principal of Immaculate Conception School (a nun whose name I have forgotten) announcing over the PA that we were to "say a prayer immediately for a grave intention." She didn't go into detail, and most of us assumed that nuclear war was beginning somewhere in the world. We prayed, and I now confess with some shame that I was relieved to find out later that it was "only" that our President had been shot. (I had seen all the cheap monster movies about the aftermath of nuclear war, and I think we forget today what a looming shadow it was back then, even—or especially—to our children.)

I can't say much about Kennedy's death that hasn't already been said countless times. I myself believe that there was no conspiracy involved, and I give no further thought to it. Kennedy was not a bad president; he brought marginal income tax rates down to sane levels (something few remember and most Democrats would prefer to forget) and made the Russians blink over Cuba. I roll my eyes at the whole Camelot fetish, though—mystique is a very dangerous thing to have in politics, and more on that I won't say for now.

It's interesting that two other great men died that very same day, including one whom I consider perhaps the most brilliant popular writer of the 20th Century: Clive Staples Lewis. (The other who died on that day was Aldous Huxley. Kennedy, Lewis, and Huxley died within six hours of one another!) Catholic apologist Peter Kreeft took advantage of that remarkable coincidence to write a peculiar but entertaining little book called Between Heaven & Hell, which postulates a drawing-room debate among the souls of these three men, somewhere away from Earth but not quite to heaven. The book is fun, even though Kreeft is clearly taking Lewis' side and makes Huxley come off less well than I think he deserved—and Kennedy somewhat better. I found it interesting to read of Kennedy's view that Christ was "man becoming God," which in some circles is what happened to Kennedy after his death. Altogether, not a bad book, and a quick read if you need one.
November 21, 2003:

In response to my question posed in my November 19, 2003 entry, several people wrote to tell me of two long-running Web diaries that go back at least to 1998, when I began the process myself. One is Jerry Pournelle's The View from Chaos Manor and the other is Robert Bruce Thompson's Daynotes Journal. Both are worth reading (though Jerry's formatting is kind of chaotic and confusing) and it's astonishing to me that both began publishing just about precisely the same time I started my VDM Diary, in June of 1998.

I got into this whole Web diary business at the behest of Lisa Marie Hafeli, who sold ads for Visual Developer Magazine right until the end. One of our advertisers (and I confess I've forgotten who) suggested that I post a short comment on the industry every day as bait to get people to come to the Coriolis Web site. It took me some time to find my voice, but I posted my first entry on June 5, 1998, and have been at it with only a single major gap between the time VDM folded and the time I kicked off ContraPositive in July 2000.

Some of the older VDM-era entries are missing, including the last two full months that I wrote it, sigh. I did not do my own HTML formatting back then; one of the Coriolis IT guys did it for me, and although I remember writing entries until the end of December 1999, November and December are gone and I can't find either the HTML or the text, even on The Wayback Machine, where access to the VDM Web site has been blocked by some joker who re-registered the domain and sells advertisers the "dead clicks" from people looking for my old magazine. There was then a six-month gap until I began writing ContraPositive, in July of 2000. That may disqualify me from being continuously in business since 1998, but hey, I think I post more text more often than almost anybody else out there. And although I grudgingly admit to writing a blog (if pressed) I really, really hate the word.
November 20, 2003:

I've been having a peculiar problem with PocoMail since I installed POPFile once again. There's an interaction between the two that I've been unable to troubleshoot: Occasional messages will pass through POPFile incompletely, and appear in PocoMail's inbox with headers intact—but no message body. It happens on virtually all messages with attachments, and in the headers I sometimes—but not always—see an error that reads: "X-Poco-Decode-Error: Cannot find boundary".

The problem occurs with PocoMail versions 2.6 and 3.0 both. Most oddly, it happens even to my two secondary email accounts that do not download mail through POPFile—which suggests that the problem is more Poco's than POPFile's. It happens only under Windows 2000; my XP lab machine receives mail perfectly even with an identical setup.

It's enough to make you nuts: I finally get a spam-management strategy that achieves 99.34% accuracy in dividing the Red Sea of my email, and something somewhere is munging messages. I'm stalemated for the moment, and if you have any sugggestions I'd love to hear them. (Upgrading my main system to XP is off the table, sorry.)
November 19, 2003:
Jeff's House Coming?
Laying kitchen tile
(102K image)
Great room tile
(61K image)

A few odd lots on a very busy Wednesday:

  • Funny how we see things that aren't—quite—there. On a box of organic high-fiber cereal that we brought home from King's Soopers the other day, a weak "o" glyph in one of the packaging fonts led me to believe that the cereal was made "from whole cats." I guess the fiber is in the whiskers.
  • Pertinent to the above: A local realtor who advertises heavily on bus benches here is named Trish Engels. But no matter how often I see her ads, I read them as "Trash Angels." That would be a great title for a story of some kind. Maybe I'll write it someday.
  • My copy of POPFile continues to learn, and as of this morning, with 5,658 messages having passed through its knowledge base, has reached a 99.36% accuracy rating. Encouragingly, nearly all of the recent misclassifications have put spam in my inbox, rather than real mail in the trash. It's been working so well I keep wondering when somthing will begin to go wrong...
  • Work on the house continues apace. We have the big garage doors in now, and there are locks on all the other doors as well. This is good, as they delivered a fortune in oak kitchen cabinets yesterday, as well as a huge pile of electrical/lighting fixtures. Much of the molding and door casing on the lower level has been installed, and all the bathrooms have been tiled. Tile on the main level is about 50% finished as of an hour ago, and should be completed by Friday night.
  • I don't know if I've mentioned it here before, but I've recently begun reading a new aggregator: TechDirt, which is a little like Slashdot only a little less nerd-specific. There's some overlap with Slashdot, but it's not 100%, and unlike Slashdot, I can generally click to the articles cited without having to wait a day for the HTTP tumult to settle down.
  • Does anyone know of any Web diaries that have been in continuous business since 1998 or before? I lost some months' worth of entries from my VDM Diary (which was hosted on the Coriolis Web server rather than my own, and the last few months were never posted!) but with a gap here and there I've been doing this since June 1998. Do let me know.

November 18, 2003:
Jeff's House Coming?
Great room ceiling
(135K image)

Our house is clipping right along; Carol and I just got back from our morning visit (we live only a mile and a half down the hill) and it was bedlam up there. Things have progressed amazingly since my last report here. To wit:

  • The interior walls have all been painted.
  • The great-room wood ceiling has been completed and stained. (See link at left.)
  • The custom-pattern porch railing has been installed.
  • The chicken-wire-and-tarpaper prep work for the stucco has been completed.
  • The subflooring has been mostly laid, in preparation for the floor tile.
  • The door casing and floor molding stock has all been stained and is stacked up in my lower-level workshop waiting for the flooring to be completed.
  • Several of the interior doors have been set.

When we left, the big garage doors were being unpacked for installation and the tile guys were hauling big bags of grout into the house. I wasn't expecting things to move this quickly, and the builder thinks we may be able to move in by January 1, which would be a full month ahead of expectations. That being the case, those of you who send us Christmas cards, hold off a bit. We're preparing our own cards here for (uncharacterically) early mailing, and our card will carry our new address. We have the key to our new mailbox up there and check it daily, so there's no reason to cling to our rental address.

I'll post a new exterior photo once the stucco is done. Most recent progress has been on the interior anyway.
November 17, 2003:

Some time back, I read an article in the Wall Street Journal about new high-tech urinals that don't use any water at all. (It may seem an odd thing to read about in the WSJ, but only if you've never followed the WSJ, especially their delightfully idiosyncratic "A Head" stories on Page One.) I grinned and thought no more about it until I actually encountered said urinals in the Colorado Springs airport the other day. The Falcon Waterfree Urinals are brilliant in their simplicity: A more or less conventional trap is created with a layer of light mineral oil that floats on urine. Urine entering the urinal drops onto the layer of oil and passes through without mixing, and the oil prevents any odors from rising out of the urine. No flushing is ever required; in fact, if any quantity of water is thrown into the urinal at one time, it will flush the oil down the drain and ruin the mechanism.

The downside is that (ladies, trust me on this) forceful delivery of urine into the urinal splashes urine droplets all over the device, and the device thus needs to be wiped down at least daily to avoid it getting, well, crusty. This increased maintenance burden is offset by the total lack of water usage, which matters a lot in arid places like Colorado Springs.

This seems like a way better method of reducing aggregate water usage than low-flush toilets, especially the sort of high-pressure, low-flush toilets that sound like a thermonuclear bomb detonating, and spray droplets of newly-minted sewage all over the bathroom. (We have one of those here at home. Loathsome thing.)

So never let it be said that you only hear about new computer technologies here. Nossir. I'm a full-service, broad-spectrum technology fanatic.
November 16, 2003:

Sorry for the long silence. It wasn't deliberate; Carol and I flew to Phoenix on the 11th for our quarterly Paraglyph Press meeting, and my USB ZIP 250 drive simply would not read the cartridge on which I keep all my Web content. The drive would read all my other cartridges, but not that one—and now that I'm back home in the Springs, my main desktop machine reads all my cartridges, including the balky one, without any trouble. Gakkh!

I am not especially happy with the performance of the ZIP 250 line. I used ZIP 100s for a number of years without any trouble at all. The 250s have always thrown me corrupted files from time to time, and now and then a cartridge will refuse to eject without rebooting. I don't know if it's the drive or the drivers, the format on the cartridges, evil spirits, or what, but these things are definitely getting on my bad side.

I'm pondering something a little more radical now: Moving my ZIP 250 cartridges to "keyfob" USB Flash drives. The 256 MB units are still about $80 - $100, but I've watched the prices on the higher-capacity units drop steadily, and it may be time to pick one up for a feasability test. I'm a little nervous about some of the reviews on Amazon indicating that the devices fail readily, so I'll be doing some serious research first. (One drive was justifiably bashed for being unable to support Windows long file names!) But boy, being able to eliminate all moving parts certainly appeals to me. I'll let you know how it goes.
November 10, 2003:

After a nice lunch today down at China Wok, I opened my fortune cookie to see the message shown at left. Sometimes there's just nothing more you can say!


November 8, 2003:
Jeff's House Coming?
Front walk and entrance
(82K image)
Our new front door
(197K image)

Whatever you do, don't buy new Belkin wireless routers. They're trying something that I consider a very bad thing: Every eight hours, they comandeer an HTTP request passing through the router and instead return an ad. As my mother would say, "The gall!" The story was first posted in The Register, though I've seen it on several different aggregators, and initially on TechDirt.

I wondered at first what these guys are smoking, but I think the answer is straightforward: They're experimenting with ad-supported hardware. In other words, in an extremely competitive market for consumer Wi-Fi gear, they're looking for any edge that would allow them to cut prices without slashing revenues too far. Ads are an odd thing for a router to deliver, but if Web sites can do it, why not?

Simple: It's not my choice, and if it's my router dictating such things to me, I have no way around it, as my only path to the Net lies through the router. To avoid running afoul of Federal law, they'll probably have to put some sort of shrink-wrap agreement on the box allowing them to basically steal Web page fetches every so often, but in truth, this is a trend I do not want to encourage.
November 7, 2003:

I am in awe. As of this morning, POPFile has processed 2,300 messages, with only 20 bad calls. That's a 99.13% accuracy rate—and it's even better than it seems, because only 7 of those 20 misclassified messages were false positives. The remaining 13 were spam tagged as legitimate, which I can deal with much more readily than a legitimate message misidentified amidst a mountain of spam. (96% of my traffic since November 1 has been spam.)

So what's the outlook for winning the spam wars? I'm more confident in technological fixes than I used to be (and still oppose legislation crafted by the sorts of idiots that politicians invariably are) and I offer an insight that I had about spam the other day, while reviewing several hundred messages to see if POPFile had tagged them correctly: Spam is targeting a relatively unsophisticated audience that is new to computing. Take out the drug spam (which appears to have replaced porn spam almost completely) and the spam I get looks more and more like the Shopping Channel every day. The ads are overwhelmingly slanted toward a low-income, socially and politically conservative audience with a high school education and no more. Spam subjects like "The Deck of Weasels" (cards backed with pictures of Iraq War opponents) and "Stop Hilary in Her Tracks!" (guess who?) are hardly the sorts of pitches that would play well with an educated, liberal coastal crowd. "Make Money at Home" pitches have always been targeted at desperate young mothers with small children at home and no career prospects. I have no idea who would ever want a replica of "JLo's Pink Engagement Ring" but I suspect Harvard grads are not high on the list.

Why is this important? Simple: Sooner or later these people will catch on and stop buying things from spammers, as they accumulate some computer experience. After another hundred thousand of them get caught in spammer swindles of various sorts (and we don't need no steenking Nigerian money scams for that!) word will get around. PCs are about to become ubiquitous, and after ubiquity comes (I suspect) a modicum of wisdom. Adding authentication to POP will make defeating spam possible, but to really nail it, we need to get people to understand that email marketing is bad, period. Over time, spam will teach that lesson itself better than any of the rest of us can. In the meantime, there's POPFile. 99.13% accuracy isn't bad—it's freaking amazing!
November 6, 2003:
Jeff's House Coming?
Standard front view
(82K image)
Smoothing the walk
(197K image)

After about ten days of very little progress, alluva sudden things are popping up at the house. Yesterday they primed all the drywall, making it a uniform blinding white—and afterward, being in there was like walking around inside a ping-pong ball. About noon today they poured the front porch, front walk, back stoop, and driveway apron. After watching them get it all troweled smooth, I had a geistesblitz: Create a jig to impress our initials and the year neatly into the concrete somewhere. So I took a coathanger and a piece of scrap 2 X 6, and shaped small pieces of coathanger wire into reverse letters, which I hot-glued to the back of the wood scrap. It took less than half an hour total, and by the time we got back up to the house (we're only a mile and change down the hill) the concrete was still soft enough for me to press the scrap and its wire pattern into the surface of the back stoop. It worked well, though the top part of the "3" came loose and didn't impress quite right.

The roof is now mostly finished, and there's probably only two more days worth of work left on it. The drywall texturers were hard at it when we left at 4 PM, and the stucco crew was sheathing the outside of the building in felt and chicken wire. Electricity and gas have been connected, and the furnace cranked up to keep the interior at a temperature that allows the wallboard mud to dry. Our Craftsman-style front door has been delivered, but won't be placed until the stucco is done. However, it's becoming amazingly like a house in a very short time. (Hard to believe we've been at it now for almost exactly four months!)
November 5, 2003:

Last week, a reader who didn't sign his name tipped me off to the presence of a copy of Alfred Morgan's Boys' Third Book of Radio and Electronics on Amazon Marketplace, for only $45. I'd never seen it for less than $100, which is way more than I'll pay for a book of this type. It arrived yesterday.

This was the main book I learned electronics from, back in sixth grade in 1963. I built the little 2-transistor amp shown at left, and it worked pretty well—or it did once I got the battery polarity correct. It allowed my 1N34A diode crystal set to feed a speaker and be heard across the room, which I thought quite a trick, until (on a hunch) I hooked the crystal set directly to the speaker and could still hear it across the room. (The house where I grew up in Chicago was looking right down the throat of 50,000-watt hillbilly rock station WJJD AM's directional antenna array, just a couple of miles away in Park Ridge.)

Morgan's book is filled with crystal-clear hand-drawn illustrations, of the type that filled the pages of Popular Electronics and Electronics Illustrated and allowed us to build radios by duplicating a drawn layout rather than reading schematic diagrams. (I learned that later on, by comparing finished radio lashups to their schematics.) This is a species of art that is pretty much dead, and neither photographs nor computer drafting seems to have anything like the same clarity and appeal. I still wonder whether he traced photographs of finished gear, or drew it all right out of his head. I think he drew them; I have books of his dating back to the 1920s, and although he might have traced photos back then, I think it more likely that he drew them from rough sketches with a damned sharp eye for perspective.

However he did it, the man was amazing: He published books over a 60-year timespan, and generations of kids (mostly boys; girls didn't build radios in the 60s) learned the pleasures of building things from his books, which spanned a lot more turf than simply radio and electronics. I have all four of the Boys' Books of Radio and Electronics now, and with that on the shelf I now have all of his work that I used to take out of the Carol Roden branch of the Chicago Public Library. Look for something long enough, and you'll find it. Lordy, we live in wonderful times!
November 4, 2003:

More odd lots on a very busy morning:

  • Pete Albrecht sent me a page of winners in Oceanside Photo & Telescope's Mars Photo Contest. There were a lot of winners; be sure to scroll down the page and see them all. People who have looked at Mars through modest-size telescopes (8"-14") will doubtless be amazed at the detail available in the photos. I think the reason they come out so well is that digital photo processing factors out atmospheric disturbance to a great extent, but in truth, I'm not really sure. These are pretty damned amazing, though.
  • On the other hand, a chap in Holland claims to have taken high-resolution photos of the surface of Titan 8" telescope. To his credit he did the math, and indicated that the disk of Saturn's largest moon is only 0.86 arc seconds in diameter, but I confess I am very skeptical. Pete Albrecht points out that if you turn enough software loose on a few pixels of light, you can generate all kinds of interesting things...
  • This morning POPFile passed the 1000-message mark, with only 14 classification errors. That's a 98.6% accuracy rating, which I find adequate in my own work. The errors were evenly divided: Half false negatives and half false positives. 1000 messages is just three days' traffic for me (and inexplicably, my message count is down about 20% in the past week or ten days) so it will be interesting to see how it fares after a full month and 10,000 messages float by.
  • Jim Mischel asked me what I thought of Novell's purchase of SUSE, Europe's most popular Linux distro company. Novell has, in the past, bought Microsoft's competitors and run them into the ground. The kicker here (and the reason I'm not too worried) is that we're talking open source now; a distro is really just that: A distribution organization. The software lives a separate life, and if Novell abandons SUSE, somebody else could pick it up. The GPL won't allow anyone to just "sit" on the software. Ditto Ximian, which Novell also bought earlier this year. I have high hopes for the Ximian Desktop, and as long as it remains open source—and as long as people like Michael Icasa retain their interest in evolving it—who actually "owns" the software is a secondary consideration. They don't own the software—just the company, and anybody can start a new company.
  • Not much new on the house. When they finally finish the roof I'll post some new photos. As I mentioned earlier, the builder took a week off, and work basically stopped.
  • Reader Roy Harvey tried Cat Bowling (see my entry for October 31, 2003) and did considerably better than my middling best score of 146. In fact Roy got to 391,771,888—most of which was achieved on the sixth frame. Talk about making it count! (And somebody really needs to take a look at that Java code...)

November 3, 2003:

The elaborate spam filters I have spent much of this year creating in PocoMail have been failing more and more, as the bulk of my spam is now arriving with forged headers and keywords interrupted with random punctuation. Spammers aren't using "V1agra" anymore; they're now sprinkling in random characters, giving me ever-changing variants like "v|ag`r-a. Back in the summer I was catching all but four or five spams every morning (out of the 200 that typically arrive during the night) and by Halloween it was up to 35 or 40.

So I did what I didn't think I would do, and downloaded the newest version of POPFile, a naive Bayesian filter for separating your mail into categories, like spam and non-spam. I experimented with POPFile late last year and early this year, and spoke much about it in these pages. It threw false positives (legitimate email tagged as spam) far more than false negatives (spam tagged as legitimate) which is bass-ackwards from what I need, so I dumped it about the time we moved here from Arizona. I installed the latest release (dated October 3, 2003) late in the evening on November 1, and I am amazed at how well it works. Since then, it has classified 582 messages, and missclassified precisely two: One false negative, and one false positive.

That's pretty damned good. It helps that I retained the Bayesian database accumulated over my first test, and it also helps that I immediately created "magnets" (keywords that "pull" messages toward one category or the other) for all the email newsletters I get, and people from whom I receive the most real email. It took ten minutes to create the magnets, and I will create a few new ones in coming days. The program isn't perfect (it really needs to treat my entire address book as a whitelist, and not force me to create a magnet for each entry) but it seems unfazed by the inclusion of "ordinary" text in spam messages (a recent spammer innovation) which I thought would render a naive Bayesian filter completely ineffective. POPFile is free and you lose nothing by trying it, and its authors are doing their best at staying ahead of spammer tricks.

So far so good. I'll let you know how things go as time passes.
November 2, 2003:

I learned today of UPN's new TV series "Jake 2.0," which is about a young man who is accidentally injected with millions of nanomachines that give him super powers. (I liked such things better when they involved radioactive spiders.) The problem, of course, is that the core premise of my novel, The Cunning Blood, is perilously close: A distributed computer consisting of billions of individual nanomachines lives in the bloodstreams of members of a secret society. The nanomachines impart certain advantages to their "operators" (nobody says "hosts") and though I wouldn't call them super powers, those advantages are quite powerful: rapid healing of wounds, immunity to diseases and poisons, short-term heightening of strength, things like that—in other words, things that don't violate the laws of physics as we understand them.

I guess it's an affront to my pride, but what I really don't want is to be accused of swiping ideas from an idiotic TV show. I worked out the premise of The Cunning Blood in 1997, and finished the novel early in 1999. With some luck I'll publish it myself next year—but I'm already afraid that the ideas I was playing with will be commonplace by the time the story gets a hearing in the SF marketplace.
November 1, 2003:
Don't buy new Symantec products. They've instituted the same sort of product activation system that Microsoft uses for Windows XP and Office 2003—in other words, you can't use the software unless you get their permission, even after you've paid for it—and the technology doesn't work. I object to product activation on a number of grounds, but the most practical is that if something goes wrong—if the company gets sold or goes under, or decides for whatever reason that it doesn't want to "activate" your copy, you have given them your money and gotten nothing for it, with no recourse except (if you're insane) to sue a billion-dollar corporation. I've been a satisfied Symantec customer for years (with Norton AV) but if they ever try to force me to move to a newer version requiring product activation, I'm going to some other product. Adobe is doing the same thing, so at some point I may have to give up InDesign, which I also like—but not so much that I'll play their ridiculous "Mother, May I?" game. (And if you're going to impose a system like that on your customers, it would be nice if it actually worked!)