Older blog entries for thull (starting at number 5)

Back from Idaho yesterday. Threw out a lot of junk email today. So it goes.

The more I find out about the home networking and automation opportunity, the more overwhelmed I feel. Dug up stats saying that the current build rate for single family new houses is 1.5M/year, with a median price of $197K. OK, that works out to $295B/year. (Actually, the $197K is only for spec homes, about 2/3 of the total, the other 1/3 being custom-built; I'm guessing the latter are more expensive.) If you could put a decent $5K system into 10% of the new house market, and a bare-bones backbone $2.5K system into another 10%, you're talking $1125M/year. We're talking about a low margin service business here, but it's still a nice opportunity. (And the World Domination advocates should appreciate the strategic value of having free software as the backbone of every modern house.)

The big problem I see is not the product -- aside from some software gaps and cost issues that volume could fix, almost everything is ready now, off the shelf -- but getting the service and market messages together before the waters gets too polluted with all the other companies who can smell this market. (Fortunately, most of these companies are assuming Microsoft software, so don't expect much from computers.)

Leaves me wondering how does one find compatible entrepreneurs out there wherever?

Packing tonight, setting out for a long car trip tomorrow, across the mountains and out to Idaho. I guess it's the vacation I never got around to taking this year. Hopefully when I get back I'll know more about how the home automation business thing shapes up.

Finished reading Donald Rosenberg's Open Source: The Unauthorized White Papers. It's a useful book, basically a compendium of open source-related business activity. Not much earthshaking news, but it seems to be comprehensive enough. One bone is that while he has a lot to say about businesses making money off of open source, he doesn't have much to say about users saving money (and otherwise leading happier lives) by using open source software. He also misses the whole freedom issue, which should be important even to someone with a fairly narrow business focus because freedom is the engine that drives economic progress.

Deep breath. Had a fairly long entry underway when X bit the dust. This doesn't do much for my faith in the legendary robustness of Linux systems.

Thrashing a lot today, as various things popped into view. I saw a notice that on contex.com that the color prepress products that I had worked eight years on have been discontinued, that support contracts will not be renewed, and that the only support contact going forward will be an ex-employee with no access to the source code.

I find myself with very strong feelings about this, partly I'm sure because it was a critical and formative period in my life which left me with very strong emotional ties to the product, my colleagues, and our customers, but largely because I always identified with those customers -- who today are being told by Barco (a competitor who bought Contex two years ago) to abandon their SGI-based CEPS systems in favor of Barco's NT-based ones.

The lessons here are the obvious ones: that engineers who write software property lose it to the whims of the property owners; and that customers who rent that property in fact own nothing and have no rights. This is, of course, something I learned painfully over many years of doing just that, both as engineer and as customer.

Other items that popped up:

  • I read with interest raph's proposal to use the trust metric to help consumers sift through a vast distributed archive of MP3 music. One issue here is that music taste varies wildly; consequently one's view of who to trust is equally personal, and that is ultimately the only criteria that matters. I'm unusual in that I make almost all of my music buying decisions based on printed reviews. I've done this for a long time, and I buy a lot, so I've accumulated a lot of data which guides me in whose recommendations I do or do not trust, and one thing that I've learned is that no one reviewer is equally good in all fields. Still, my experience is with named writers, necessarily a small set; I wonder whether larger reviewer sets might start to become more reliable?
  • Got mail today saying that sourceforge will be using the trust metric to allow everyone to rate anyone. This strikes me as a pointless game -- although I'm sure that anyone who gets identified will be quite deserving. I suspect that one problem is that what the metric may turn out to measure is the topography of the developer community -- basically how closely groups of people work together, because the larger the population the less we really know about each other. The same dynamic may be present at Advogato, but the relative crudeness of the rating scale should make it less pronounced.
  • I understand that Sun has some scheme that allows one to compile Linux drivers and load them into Solaris. Fwiw, I urged SCO to do this same thing. My proposal was rejected, because the powers-that-be felt that allowing Linux drivers to be used in UnixWare would diminish the message that UDI is really the way to go.

The basketball goal took priority over the resume, but I finally cleaned up the resume and posted it here. I meant to hang hypernotes all over it, but was finding I could write forever on them, and it just gets more and more scathing and/or pathetic. Some day I'll get around to those notes; some day I'll call it autobiography, and it will be funny.

I also jotted down a couple of project ideas:

  • A database for recorded music. (The fruits of endless aggravation with the All-Music Guide.)
  • A business plan based on integrating a Linux firewall with structured wiring and home automation gadgets, and building this into new houses.
Needless to say, anyone interested in such projects please get in touch with me.

Working on resume. Last one I wrote (approx. 3 years ago) was something like 10 pages long, mostly because I felt some need to explain the context in which I did all of those strange things. I figured that the more people could read and know, the fewer dumb questions I'd have to fend, the fewer pointless interviews, etc. Of course, no more than 3-5 people actually read the thing, but they were entertained, and I did get a job (albeit not the one I was looking for). So this time the plan is to hypertext it: short summary with links to juicy details. Maybe I'll get it out today; or maybe I'll get my new basketball goal set up.

I was reading Andrew Leonard's Salon piece on how IBM wised up to open source, and saw a link to an old essay by Richard Gabriel, called "The Rise of Worse-Is-Better". Gabriel talks about two approaches:

  • The MIT Approach, characterized by the phrase "the right thing", which aspires to be correct, consistent, and complete.
  • The NJ Approach, which favors simplicity over consistency and completeness. (A better name for this might be "Simple uber Alles", or even KISS.)
The core argument is that simple systems are more accessible, more adaptable, have better survival characteristics, and therefore proliferate widely, whereas "right thing" systems are harder to build and maintain, are more expensive, etc. This much is pretty straightforward, and plenty of examples pop to mind. However, the interesting point is the assertion that people willingly adapt to the simpler systems rather than waiting for the "right thing" systems to adapt to them.

We see evidence of this all the time, but it's hard to shake the conviction that "better" must really be better. I used to work in the typesetting industry, and one of the things I worked on there was trying to automate the aesthetic rules of fine advertising typography -- kerning, hung punctuation, river avoidance, staggered rags, etc. -- but in the long run such concerns turned out to be irrelevant. It turned out that desktop publishers were so happy just to get their pages instantly, saving them trekking to the type shop and paying out a small fortune, that they were willing to forego a lot of finery.

But the arguments persist, ad infinitum, and they're hard to settle -- partly because nobody really argues for worse, the winners of "worse-is-better" just do it.

Last week (Thursday 7 Sept 2000) SCO laid off 190 employees. I was one of them.

The layoff was in preparation for finalizing the acquisition by Caldera of SCO's Server Software and Professional Services divisions. The layoffs should save the Caldera something like $5-7M/Q, which given that SCO's non-Tarantella divisions lost $10M last Q, and that Caldera itself lost some $7M in its last Q, isn't enough to make the new combine whole, but is a start.

The company line is that the people who were laid off were "redundant", and certainly there was some of that here. But there may also be some sort of behind-the-scenes battle for the soul of the new company. Certainly, one reason that SCO has been losing money all year is that it has faced ever stiffer competition from Linux, and that they are selling out to a Linux company can be viewed as capitulation. However, the Linux company in question (Caldera) is only generating $1.2M/Q in revenues, whereas SCO's Server/Services groups still account for $25M/Q. Even with the layoffs (and I've heard that Caldera laid people off, too, but not how many), SCO will still account for 75% or more of the combine's employees.

What may have made me "redundant" was that I was one of the few SCO employees working on open source projects -- specifically, the long-promised Linux port of sar (which also got shelved last week). SCO tended to view such projects as good will generators, but it's not too hard to imagine that the SCO managers who drew up the pink slip lists now see that as an unnecessary luxury since Caldera already enjoys all the good will it needs.

Or it may have just been a mercy killing: I had long argued that SCO's proprietary OS business was a dead-end, and that they had to move aggressively into Linux; that the real way to build a Linux business is through service, not proprietary bundling; and that what little value UnixWare still has is as historical legacy. None of these efforts amounted to a thing (other than possibly wearing my welcome out). It's been the most frustrating thing I've ever attempted to do, and I should be glad to put it behind me. (Keep telling myself that.)

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!