Ion is very interesting indeed... a window manager for X that's almost entirely keyboard controllable, so you're only as mouse-bound as the applications you run. I have my doubts about some of the details of how it's done here, but basically it's very nice... I may even keep using it.
09:19:51 PM, Wednesday 14 March 2001
The classic Slashdot question "Have you read the Moderator Guidelines yet?" has never felt so meaningful as it did just now, after moderating up a dirty physics limerick.
05:34:39 PM, Wednesday 14 March 2001
Three thousand hits, Kerne? And without even a link to your site listed in AltaVista? Hmm... maybe there's something to this whole 'useful' concept.
 Except one link to your old site at toast.net... to ports.txt, of course.
12:27:36 PM, Tuesday 13 March 2001
I never would have thought it would be so fun to look through my server logs! The part that really surprises me is that I can actually recognize most of the hosts I'm getting hits from: various computers at St. John's in Santa Fe or Annapolis, @home users in Corvallis, Oregon and Berlin, Connecticut, someone with a cable modem connection in what appears to be Kansas City, someone from Steve Jackson Games--once he even followed a link from his own site's referrer log--and then, of course, there are the periodic hits from the googlebot. But even more intruiging are the ones I don't recognize immediately: someone using psi.net in Annapolis (a Johnny? it must be, but who?), someone in Japan (!) who's connected through ccn.ne.jp (looks to be a cable modem provider, but alas I can't read it), someone at Oregon State University, someone connected through proaxis.net (apparently a DSL provider, possibly in California?), someone from lmf.net (Lebanon MobileFone? eh?), another cable modem user somewhere (rr.com, like Neil, but I'm not sure where... New York City, maybe? nycap.rr.com... hmm...), someone connected through iserv.net ("West Michigan's premier internet provider")... last month, someone with an internet connection from dialsprint.net, and an @home user in San Diego... that appears to be it, apart from a search engine or two. So my question is: who is everyone? Do I know you, or did you just come across my site at random? Introduce yourselves, either through the respond link or by email. That's not a demand, of course--personally, I'd be too shy to respond to something like this--but I'm terribly curious.
05:49:25 PM, Monday 12 March 2001
What's good about items?
It's a chance for me to write longer, more polished pieces, and still have them fall under one general heading. It's a place for me to explore some of the things that I think about constantly. With some work, it might actually be made useful to people other than me.
What's bad about items?
It conflates personal narrative with discourse on general principles, and does so in a way that makes it difficult to do either one properly. I could imagine something that would combine the two of them well, but so far, this ain't it. It's hard to add things to--quite apart from the fact that the software's sealed off on emma, my laptop with the broken ethernet card, the XML interface is a bit clunky at best, and the conversion program is basically unmaintainable crap. Links between sections aren't nearly dynamic enough, outside links aren't fluid enough, other references aren't structured enough, and are too hard to do. The interface won't get out of your way (or, get out of my way, rather). If it's not trivially simple to add things and edit things, then I won't do it--that's why I made bloglet. Adding something as large as an item should require thought, but it shouldn't require more thought than is necessary just to compose the item. The interface should encourage material to improve over time.
What am I going to do about it?
First off, I'm going to rewrite the software side of it from scratch, probably in Java. It'll have a web-based interface--that's the easiest way to assemble a clean UI quickly--and it'll determine as much information as possible automatically. Internally, it will use either XML or MySQL to store the items--it may use a combination of both. It will format with a template, so that the look is adjustable. It will incorporate some kind of user response system, but I'm not sure what. And, finally, it will be divided into two different projects. And, with any luck at all, it will do all this tonight.
04:24:19 PM, Monday 12 March 2001
Kerne: you're not the only one. I removed bloglet from my .login script, and started just calling it whenever I had something to say, a few months ago. Especially for people with login habits like ours, this is often a better way to get consistently good entries. Though having it in my .login for so long helped me get in the habit of adding to it regularly, which I doubt I could have done otherwise.
03:18:05 PM, Monday 12 March 2001
While I forgot to mention it at the time, bloglet turned one year old on Friday. One year, and I'm still actually posting to it regularly. Evidently the extremely simple interface had the desired effect. (You can still see the original here, though, sadly, the first few entries were lost by a bug in version 1.0 of the bloglet program.)
02:27:00 PM, Monday 12 March 2001
Hurray! Kerne has finally gotten himself a bloglet! Most exceeding cool, say I... go check it out.
01:11:25 AM, Monday 12 March 2001
It's here at last!
10:44:46 PM, Sunday 11 March 2001
Ah, the joys of randomly browsing around the net: "Abscesses, however, should always be coyly veiled in filmy black fabric."
11:22:37 PM, Saturday 10 March 2001
Neil and Mirabai, I've made a slight change to your bloglet configuration files--it shouldn't be anything visible, it's just to get the comment system working in Lynx.
12:19:26 AM, Saturday 10 March 2001
The reason Linux isn't easy for new users is that we really don't care about new users. Free software isn't developed out of altruism. We build programs because we need them, and we build them like we need them, and we have no reason whatever to make them useful for others. If you want something nice and user friendly, something that won't require you to understand your computer, then I'm sure you can pay someone to build it for you, if they haven't already. I've certainly got nothing against non-programmers, I just can't understand why some Linux geeks want to waste time making a good system user-friendly.
04:19:42 PM, Thursday 8 March 2001
And of course, I got the link there from kottke.org, but you probably already know about kottke.org.
12:22:31 PM, Thursday 8 March 2001
Anyone out there read Japanese?
12:20:22 PM, Thursday 8 March 2001
search phrase: "turning off all security in unix"
First result: Improving the security of your Unix system"
Evidently running a system with no security at all isn't a terribly popular idea in the unix world.
07:52:31 PM, Monday 5 March 2001
If you ever find yourself with some free time and an old pair of shoes you don't plan to wear again, try taking the shoes apart. It's very educational. Now I want to make my own.
04:41:06 AM, Sunday 4 March 2001
One particularly choice excerpt from the Kiersey test: "Are you the kind of person who: (a) is rather talkative, or (b) doesn't miss much?" (Gee, are those my only options?) (This one, by the way, tells me that I'm an "idealist". And the Libertarians, of course, still believe I'm a Left-Liberal.)
12:35:38 AM, Sunday 4 March 2001
That is: I really don't think we needed a big long test to discover that I am somewhat left leaning, am not religious, and tend to resent authority, nor to tell us that Neil and Bridgie are also leftists, but are a bit more concerned and a bit more hip than I am. ;)
Still, it was amusing. As long as people are taking tests, don't miss the disgracefully biased World's Smallest Political Quiz (I'm offering a reward of 100 points to anyone who can be declared a fascist using only relatively honest answers!), and the Keirsey Temperament Sorter. The former makes you answer 10 questions and then tells you you want to vote Libertarian, the latter gives you a variation on the Myers Briggs test and then offers you vague meaningless praise.
12:23:31 AM, Sunday 4 March 2001
Well, I've taken the silly test. It reports that I am an "autonomous post-materialist". It's an ugly phrase, but better than some they offer. Their analysis of the meaning of this is, of course, absolute bullshit: my "Words to Live By" evidently include "It's my life" (I mean, yes, this is technically true, but--who else's life would it be?), "Do your own thing" (No! You have duties! Responsibilities! Besides which that just sounds stupid!), and "Image is nothing" (thirst is everything. Obey your thirst. Spr-autonomous-post-materialist-ite.). My "Icons" include "Dot-com millionaires (any of them)" (EVIL!), "Computer hacker Mafiaboy" (never heard of him, but I'll bet you money he's a cracker, not a hacker. Right. My icon the skript kiddie. i 0\/\/n j00), "Animated character Bart Simpson" (... okay, Bart's pretty cool), "Steve Jobs" (when I grow up, I want to have a reality distortion field just like Steve!), "Basketball star Dennis Rodman" (... well, okay, he's cool too, but an icon?), "Animated characted Eric Cartman" (WHAT? Listen, I don't have to take this shit from a computer!), and [drumroll please] "Shawn Fanning" (some kid who runs an mp3z site). I'm told I spend my money on "Experiences, not stuff", which, if I do, I must say is extremely foolish of me, because experiences are free anyway, and stuff, once you have it, lasts longer. I do like the suggestion that I should steal from The Man.
11:45:03 PM, Saturday 3 March 2001
So the whole thing basically breaks down into four components (or at least, four types of components):
1) A tool to convert an unstructured datastream into structured data in XML.
2) A tool to convert diverse types of XML data into a single common format.
3) A tool to harvest collections of data in the common format, based on various search rules.
4) A tool to display a given collection of data.
Each type of tool requires different sorts of cleverness. (1) has to be able to parse something that may not be in any clear format--a good example of a level 1 component would be a tool to extract articles from a news website by recognizing characteristic html for the beginning and end of the article. Probably best accomplished by handwriting a new component for each data source. (2) has to be able to recognize what tags in the input format represent what data in the common format. Probably requires an individual tool for each format output by level 1, so it would be best to tune level 1 to produce standard formats for classes of data (news article, for example). (3) Ought to be fairly basic, which is to say, the real work for level 3 should be handled in levels 1 and 2... the first two levels should already have ensured that each news article has a unique ID, that every data source has a created and modified date, that a list of keywords is available for each source, etc. (4) should again be reasonably basic--just needs to read and format the incoming xml data. The main reason to keep level 4 as a separate component is that it will then be replacable--I can have one level 4 tool to print a newspaper, another to prepare a Flash website, and so on.
The most interesting problem I've glossed over above is keyword searching, and more generally the effort to extract really informative metadata from a source that makes no effort to provide it. My instinct is that level 1 should only structure the data it's given (including things that may not be explicitly stated, like created and modified dates), while level 2 should actually try to analyze that data where appropriate. That way it should be easy to build the level 1 tool, and the level 2 tool can grow more advanced over time. Also, analysis will typically be the same for sources of the same type--keyword extraction from article text, etc. Designing a good common format could be very tricky. In the end, the common format might be best implemented as an extension of XHTML. Possibly the common format would be best implemented as a pair of files, one for content, one for meta-data. Must remember to look up the relevant w3c recommendations. Output to a really standard format would be most desirable. I think I'll try to have a component from each level working by the end of the weekend--wish me luck.
05:17:51 PM, Friday 2 March 2001
Here's an intruiging use of the web: a hypertext version of Wittgenstein's Tractatus Logico-Philosophicus. Starts out with just the basic argument in seven steps, expand as necessary. I'd seen it a while ago, but I'd forgotten about it until just now.
01:02:45 PM, Friday 2 March 2001
When trying to solve a problem, a programmer will often begin by generalizing it to the point that the solution will continue to be useful in the future. Perl is just such a generalization, except that it is generalized so far that it now applies to the class of all problems.
03:24:15 PM, Thursday 1 March 2001
Mirabai's bloglet is back... I think this is the last one I'll need to migrate over here from whorfin, for a while at least. ;)
04:22:42 PM, Wednesday 28 February 2001
"In Xanadu did Kubla Khan A stately Pleasure Dome decree..."
The most ambitious hypertext project ever released its source code to the world more than a year ago, and I hadn't noticed until now.
06:51:33 PM, Tuesday 27 February 2001
SOCRATES WAS GO DOWN TO PIRAEUS
SOCRATES: WHAT HAPPEN?
GLAUCON: SOMEONE SET US UP THE GREEK FIRE
ARISTON: SERVANT COME
SERVANT: THRASYMACHUS DESIRE YOU TO WAIT
SOCRATES: IT'S YOU!!
THRASYMACHUS: HOW ARE YOU O KALOI K'AGATHOI!!
THRASYMACHUS: ALL YOUR EIDE ARE BELONG TO US
THRASYMACHUS: YOU ARE ON THE WAY TO HAVE ARGUMENT MAKED THE WEAKER
SOCRATES: WHAT YOU SAY?
THRASYMACHUS: YOU HAVE NO CHANCE TO CONVINCE MAKE YOUR TIME
THRASYMACHUS: HA HA HA ...
SOCRATES: TAKE OFF EVERY 'MATHEMATON'!!
SOCRATES: UNDERSTAND 'MATHEMATON'.
SOCRATES: FOR MEGALODIKAIA.
04:37:35 PM, Tuesday 27 February 2001
Advogato, "The Free Developer's Advocate", is a very nice site for free software developers. It's a very good site, and I'd heartily recommend it to anyone involved in developing free software--or any software, really. Badvogato, on the other hand, is an entirely different creature, a web-based meeting ground for evil overlords and very-naughty overlords-in-training. I'd heartily recommend it to anyone, and especially to the sort of people that are fans or authors of Return of the Villains--which I'd also recommend to anyone, so there. Anyway, check out Badvogato. For best results, browse through Advogato a bit first to get a feel for the original.
09:33:27 PM, Monday 26 February 2001