Table of Contents

Small Test Entry

Hi there. Welcome to this thing.

by sbp on #sbp at 2003-10-28T19:02:46Z

Possible Features To Add

1) We could allow dogcow to use (title <URI>) or something LISPy like that. 2) How about moin moin style [URI title] links? Will require more URI sniffing, but it's not bad. 3) I think I'd like @title on the <<< line, so like "<<< Possible Features to Add" would've started this post.

by sbp on #sbp at 2003-10-28T19:52:08Z

Files That Are Commands

I was just thinking that it'd be nice if you could make a filename a link to a command that the STDOUT from is the file's content. Hmm. I wonder if there's a way to do that with shell scriptery? The problem is that you'd have to execute the file to get the content. I want the content any time you actually access the file. Which, of course, is extremely dangerous since you don't want all manner of scripts running when you merely access a file... But then it'd be powerful if you want to mirror a file but just change one small detail in it, etc., without having to make sure that the two are always kept up-to-date. So that's my usecase: I have a file xyz with "xyz a" in it, and another file blargh with "xyz b" in it, and I just want to create blargh from xyz using sed "s/a/b/", and then maintain xyz as the primary file to edit to.

by sbp on #sbp at 2003-10-28T20:16:31Z


This site is mainly powered from IRC, using a bot called blogbot. So here, now that we've got a nice new style, I thought I ought to try out some of the bot's features—e.g. automatic Unicode handling and seperate paragraphs.

If there's a gap above this line, then that worked. We should also be able to do random links quite easily.

by sbp on #noets at 2003-10-30T03:20:14Z

About the nœts Current Design

HTML design these days is quite irritating. A mess of CSS vs. tables, HTML vs. XHTML, JPG/GIF vs. PNG and so forth. What does one do? If the ViewSourceClan are to be believed, one simply copies the source of a site that one likes.

But this style was formed by accident. The top left corner image is just a standard GIMP tool (sparks) working its magic, and didn't take much airbrushing or compression to get to its current state. The colour used is just a set of random nice warm colours; winter's coming up, and you want your sites to be homely. Even the fonts are just whatever looks good. But compared to my previous "hacked-up" style attempt, I'm quite pleased with this.

Still not sure about that by-line, though. Oh, yeah, and thanks to d8uv for the translucent PNG on the entries. That's probably the coolest bit! A "recent cool evolution", as he says.

by sbp on #sbp at 2003-10-30T03:26:52Z


The Freenode IRC server today was invaded by a bot from the University of Helsinki. Called irchiver, the bot was to gather data for an IRC search engine project. A research scientist connected with a project joined Freenode to try to explain what the project is about, given the sparse documentation and general irritation felt by the Freenode users.

In general, the project has no official documentation and does not appear well thought through. But the whole thing raises many IRC privacy issues. IRC privacy is perhaps even more complex a case than with HTTP. robots.txt is a norm that IRC has no equivalent of.

I quite like the idea of being able to Google for channels, though. I still use Google to search #swhack logs with some frequency. And there's an interesting point therein: is Googling for HTTP logs of IRC good enough? An opt-in IRC googlebot would be very cool.

by sbp on #sbp at 2003-10-30T05:01:49Z

The Case of the Missing DOCTYPE

Though this is subject to change, nœts is being served as XHTML content but without a doctype declaration at the top. Validation is very useful for authors, and should be a part of the authoring process. It can show you whether or not you have errors that may impede browser parsing and display, e.g. unquoted ampersands and so forth, so it's just as important as checking the content in browsers.

But why is validation useful on the client side? To most HTML UAs, the type of HTML is unimportant. Even for those with standards compliance modes, there's always a fallback. If you want to know whether content send as text/html is SGML or XML, you can check for the presence of an xmlns attribute on the html element. It must be there for all mandated XHTMLs, and mustn't be there for all mandated HTMLs.

Along these lines is Hixie's Sending XHTML as text/html Considered Harmful, which he recently updated, and my SGML isn't widely supported reply. Dan Connolly has said a couple of times that "surely <html xmns=""> is enough". I don't think there's been a good argument against so far.

by sbp on #noets at 2003-10-30T21:05:19Z

sbp is wrong.

Noets should be more or less forced to use application/xhtml+xml. You see, if we don't, Hixie will crush us with his superior crushing powers. You see, HTML UA's pay special attention to the doctype. What it does with the doctype, I'm not sure. But I'm sure it does something important. And besides, relying on <html xmlns=""> is dumb. I mean, who ever uses xmlns? It's not needed at all. Even Tim Bray said once "xmlns is dumb, and people who use it are dumbys."* And that's good enough for me.

*May or may not be Tim Bray.

by d8uv on at 2003-10-31T01:59:44Z

PNG Translucencies & Browser Lag

The items here on nœts are styled with pastel translucent backgrounds, which is achieved using PNGs (courtesy of d8uv) with an opacity of roughly 60%. Whilst the effect looks cool, originally we were using a 1x1 pixel PNG which slowed the rendering time down—especially when scrolling—quite considerably. It took a little while to figure out that it was the PNG size and not some CSS or Mozilla-related weirdness, and once we did the situation was "fixed" to some degree by using a 256x256 pixel PNG instead.

The rendering lag is still pretty high, but at least it's managable now. If you have another solution, or comment, feel free to drop into #sbp on Freenode and let us know. Thanks.

by sbp on #noets at 2003-11-01T04:25:45Z

...Or so I wonder...

Now what will Blogbot make of this...

by xover on #sbp at 2003-11-01T20:40:21Z

Blogbot Command Syntax: VIrc...

Perhaps Blogbot syntax should be expanded to be more reminicent of "vi"?

Since "::" is treated like a comment, the colon is allready established as "magical. That immediately puts me in mind of vi's ":w" (to replace ">>>") and :q (for "@quit"). Using ":t <title>" for "@title" may be going over the top (and lacks precendence from vi), but might be worthwhile for consistency and completeness.

by xover on #sbp at 2003-11-01T20:46:41Z

Filesystem Hierarchy Standard for ~/

Ever noticed how the FHS simply gives up when it comes to talking about /home? Clearly, one would think that this is because the structuring in a user's home directory is up to them, and that even the location of /home on a system (and its existence) is optional. There are two reasons why this should change: firstly, everyone has organization nightmares; secondly, some of those nightmares are inspired by various programs eating up ~/.program_name

If you take the latter problem, this mirrors to some degree the philosophy of the FHS—though it's a bad part of the philosophy. The system gets /, and users are forced down into /home/theirname. Programs get /home/username/.programname, and as a result users get cluttered ~/ directories. Didn't the *nix developers hear about Feng Shui? Didn't anybody really think about how best to organize things? The FHS seems like a mishmash of silly conventions that *nix has accrued over the decades.

So the first thing to do is give the hierarchy back to the users. .program_name files could use the Java packaging philosophy: for example, emacs config files could be placed in ~/conf/org/gnu/emacs This still takes up ~/conf, so it's not perfect, but it's better. It also begs the question: why put important config files in a place where they're less accessible? Well, you can always make a soft link, e.g. $(cd ~/conf; ln ~/conf/org/gnu/emacs/.emacs .emacs), that you can use but that the programs must not.

Aside from giving the space back to the user, we also have to work out how best one should utilise that space. Everyone has different needs and opinions, so a tips-guide or cookbook would probably be better than a standard. There are still some technical extensions that could be made to filesystems to facilitate this, though. For example, why use hierarchies at all? Keyword based systems are favoured by some, so people should ahve the option of using them. Systems such as Longhorn's new FS, and GNOME storage are pioneering this, but hierarchy-free systems also have their limitations, so they shouldn't impose themselves on users that don't want them.

ah, good idea One other musing: why not seperate system data from user data entirely using a different prefix? I suppose that's what the ~/ syntax is for, but it could go further.

by sbp on #sbp at 2003-11-01T20:49:01Z

Improving Blogbot, and Editing From IRC

Further to xover's comments, he's come up with a few more suggestions. For a start, having an :s/from/to/ syntax would be extremely helpful, since I keep having to dip into the server to edit posts that I've screwed up on. Note that my accidentally uncommented "<sbp> ah, good idea" in the previous post made its way in (well, it'll probably be gone by the time you read this). I'm sure that could be resolved with a bit of practice... but people are bound to make mistakes, see that they've made mistakes, and want to fix them.

xover also suggests ":<something> <n>" to retrieve paragraph n for editing or replacement, as well as using NickServ as an authentication mechanism to enable people to edit or augment their old posts.

In fact, part of the background to blogbot is an idea of mine to use IRC as an editor. I was planning upon much richer editing features than I've got in blogbot, though I figured that it'd be primarily useful on a private channel, since you'd probably be recalling old test a lot. If you stuck to one line per sentence, perhaps with Nelson-esque automatic capitalization, however, you could probably even employ some sort of tumbler system to keep things in check. Then you could reference old lines by their static tumbler addresses.

Note that this'd also be a nice approach for editing in a command line session without curses. Though why anyone would want to do that, I don't really know. Perhaps conversational editing holds some advantages. It does make for some rather odd editing, however: since IRC is such a different medium to what one expects for editing, posts can tend to be a little jerky, losing the implied ,/. pause between lines. Perhaps we need a new punctuation mark for it!

by sbp on #sbp at 2003-11-01T20:58:51Z

More on Organization

I just wrote a utility script that I have the command $(nftp) for which does various FTP related functions for me. I write a lot of little scripts which glue my computer together, and, like Morbus, I'd like to release some of them so that others can use them. Most of them just do little tasks, but it's often not worth re-inventing the wheel if someone's already done it fairly competently (not that I do). Perhaps unlike Morbus, however, I have released a few. They're just not in any particular place. Some are scattered in www-archive, some are on, and some are even on miscoranda in blog entries. Yuck. In fact, I don't even have a good organizational scheme for the utilities on my hard drive. They're split across partitions, across directories. Some are on my $PATH, and some aren't. They're normally in Python, but some are in bash and so forth. It's a mess.

So, how to resolve it? I'm not sure. I tried with my u script to give some control over where the utilities are. In fact, that $(u) script has grown into a pretty large bash script now, which is yet another thing that I ought to put online somewhere. But I'm not using the u scheme as much as I used to, since I'm becoming quite used to just bunging stuff into ~/bin, which I have on $PATH.

I did wonder about creating a new Website that is basically a mirror of my hard drive. I'd set a plan that would say which fiels I consider public, and then it'd just mirror each of those files to the Web. In a kind of rsync/cvs style. To prevent annoyances from when I upgrade hard drives and so forth, I could give each partition and hard drive a name, and then work from there. The downside to all that is that a) it's a stupid plan b) it'd be a major bit of coding, and would involve buying a new domain, all just so that you can check out my utility scripts, .bashrc, init.el and so forth.

One can get quite used to crappy organization schemes, anyway. Sometimes, disorganization is the best form of organization...

by sbp on #sbp at 2003-11-02T00:26:10Z


The man page for tsort(1) really sucks, but when I read a better explanation it started making sense: it's about dependencies. In fact, there's a nice diagram of what tsort calls a partial order (a directed, acyclic graph) in the topological sorting algorithm page, with its linearization next to it, which clearly shows a dependency aspect. I found a nice implementation of tsort in Python, too.

This made me think about writing a function linearizer for Python, taking in a module and getting the code as strings for each of the functions declared within, and checking which functions are called... but then I realised that with any sort of conditionality, it all becomes too complicated for a medium-quick hack. So oh well; tsort's still quite nice.

by sbp on #sbp at 2003-11-02T05:13:07Z

Substitutions in Blogbot

Now blogbot can do substitutions. Yay! This all arose from the fact that people aren't perfect, especially when it comes to typing in prose to blogbot on IRC.

The substitution format is: ":s/from/to/[g]" where the optional g flag indicates to replace all of the occurances; only the first item will be replaced otherwise. This implements the change first suggested in Improving Blogbot, and Editing From IRC.

by sbp on #sbp at 2003-11-03T01:16:17Z

Translucency? Who Needs It?

As previously reported, translucencies render very slowly in Mozilla when used as CSS-attached backgrounds. Terje Bless has, however, come up with a modified CSS style for noets based on some translucency experiments by Eric Meyer. The "trick" to it is that there are no translucent PNGs being used at all; instead, since the top left part of the background is the only region with any texture, that's been copied and then washed out to make it look as if it's been viewed through a translucent layer. Then, the washed out image is set as the background for the entries, fixed to a position of (0, 0), which is relative to the viewport and not the entries.

References: Eric Meyer's complexspiral technique.

by sbp on #sbp at 2003-11-03T01:24:24Z

More Blogbottish Qualms

And here I go into editing mode again. Wasn't it Alan Kay that said "don't mode me"? Whoever it was, they were right—having modes is irritating, as is having syntax that does not come naturally to the environment that you're working in.

And so it is with editing these noets entries on IRC. One thing that I'm noticing right now is that people are talking to me whilst I'm trying to come up with a post that encapsulates what we were discussing five minutes ago. So the conversation moves quicker than the publisher. But what we were talking about five minutes ago is that, as noted previously, my entries are jerky and don't come across well. What do you expect? I'm in a conversational-state interface here, not an editor. It's line mode.

Another point: I'm much slower writing these sorts of entries than I am when I'm just chatting. That's because I'm thinking about the sort of content that I want in the post, and how to word it. Since I'm publishing this to the general public, I want to keep the quality higher... and that means that I slow down, and often lose the thread of what I'm saying. It's harder to skip backwards when you're entering large lines, too—you have to remember what you said instead of being able to go back and read it.

Which leads me to the conclusion that perhaps it wasn't such a good idea to set up such an editor from IRC. It is, however, nicer than editing from a Web interface in that I've clearly made a lot more posts to noets than I have to miscoranda... so in a way it's been a win. Just not a clear enough win to make me sure that continuing with this format is a good idea. For example, I'm having to redo this post in #noets because I got disconnected during my original posting of it in #sbp, and came back on and forgot I was editing, destroying the post with IRC-talk. Oh well.

by sbp on #noets at 2003-11-03T01:49:02Z

about content

I do have three things to say about some content issues of this fine site. The first one is that almost all of the posts on noets (including this one) are about noets. I can handle metaness as much as the next guy, but this is getting insane. From feature announcements to changes in the CSS (BTW I added the Owen Hack so IE doesn't blow up the complexspiral translucency), this site is getting kinda unruly with self-referentialism.

If we don't write about ourselves, then what do we write about? Is it a collective weblog about the lives of its contributors, or should it be more focused on a topic? And if so, what topic? This should be addressed sooner or later, while still keeping in mind point #1.

My final point has to do with the awkwardness of writing weblog entries in IRC. I will admit it is awkward, but mainly because you're basically writing in a different type of medium. It's kind of like the difference between AM and FM radio. They both are transmitted using radio waves, and turned into sound using basically the same process. But the fact that most music channels are in the FM spectrum, and most talk channels are in AM, shows that there is an underlying difference between the two. That difference is the quality of the final broadcast. In weblog entries and IRC chattings, however, the difference is the length of the editing window. In HTML forms, you are given a large box in which to compose, and a submit button once you are done with the whole thing. In IRC, however, you are given a box that only accommodates one sentence at most, and a submit button once you are done with that sentence. But, overall, I don't think this will become a major issue. Although it's "jerky", it's not too bad. Besides, most newspapers are written a little "jerky" and "choppy".

by d8uv on at 2003-11-03T09:56:57Z

Mail Archive

Random find of the day: The Mail Archive. All you do is subscribe their service to your mailing list, and it automatically archives it for you. Although there are no guarantees about archive persistence, it does seem to have been running since 1998—why did I only just find out about this? And it was by pure chance: I middle clicked a highlighted URI in Mozilla Firebird to see how it would resolve, and it ended up searching for me.

They even support RDF. Now if there were just a reliable service for providing mailing list administration, it'd be very easy to create a new list.

by sbp on #sbp at 2003-11-03T20:59:51Z

More Noetsy Self-Referentialism

As noted previously, the majority of posts on noets so far have been about... noets. That's fine. Let's expand a little on what the intended scope of noets is: whatever. It's for whatever project is current, whatever news people pick up from elsewhere that they want to valuably comment on, and so on. The current project that I'm finding quite interesting in its technical challenges is this very site—once it's all stable and we've been using it for a while, I'm sure that the content will shift away from noets related issues; there can only be so much about this site that we can write about!

You can take the post on tsort as being most indicative of the future direction. I'd like this to be a cross between weblogs I've been involved with (miscoranda, #swhack, TNS), taking their best bits. Those weblogs are mainly oriented to the kinds of things that we talked about on #swhack, and talk about on #sbp, hence it made sense to run blogbot in #sbp and closely related channels.

I note that I'm referencing previous posts quite frequently—perhaps there's a need for threading here? And when I was looking up the references to the weblogs I've been involved with, I rather lost the thread of what I was saying...

I was also hoping that noets might become a place for communal participation, which extends over HTTP, IRC, and email. That's why the publishing process (blogbot, basically) is open to anyone, and it's also why I'm considering allowing vanilla comments on the site. As to email, I'm not sure how I can involve people there yet.

But again, as I'm hopefully already on the record as saying, noets is very useful in getting me to record various technical materials, and if you compare its output to the previous blogs that I've run, it compares quite favourably. Oh, hey, except with BIOH which was rather fun.

by sbp on #noets at 2003-11-04T00:08:50Z

Eddie, Google, and Logging IRC

Are Google Crawling IRC? (via Google Weblog). These bots have been running for a couple of weeks now, always with eddieNN usernames (where NN is a two-digit number) and Google IPs. Given the recent irchiver situation as well, it seems that this may be the season for search engines running bots all over IRC!

It's difficult to speculate what the Google bots are doing, and the behaviour hasn't been widely reported yet—the only other site so far being Since the bots aren't staying around for long, people have posited that they're getting user/topic/spambot URI information... but that wouldn't seem all that useful. In fact, sites such as and have already proven that there's not much information value in IRC administrivia.

Google claim, apparently, that the "behavior is only temporary". The other interesting detail is that the bots have been crawling small unimportant networks. As noted on the post: "Today eddie08 was observed on a really small non-announced network with only one channel, is it our google overlords trying to do some social network experiments?" We've had another independent report saying the same thing—small network, but Google bots. So perhaps Google are trying to provide better functionality for people searching for obscure IRC channels that could be mentioned in any one of a number of ways in Google's current datastock?

Curiously, the bot's realname is given as "walso". I doubt that the eddie/walso combination gives anything away as to the bot's purpose, but that's another interesting point—why not choose a name more fitting of a bot rather than a person?

by sbp on #sbp at 2003-11-05T00:20:10Z

wow, this is neat

by Ash on #sbp at 2003-11-05T00:41:12Z

Words of Choice

I guess it should come as no surprise to anyone that lexicographers are often asked what their favourite words are, but perhaps it's more surprising when one of the most esteemed publishes a list of his own words of choice that are both common and rich in the delights to be yielded from their etymologies.

John Simpson, chief editor of the OED, presents ten words and phrases noting that really lexicographers don't have favourites. I don't believe him, though: in the OED newsletter that I got this from, the newsletter's editor polled their collegues and came up with another more standard list of "favourite" words: but, buttle, checkmate, chocolate, delphinestrian, gazebo, lucubrate, mumpsimus, persiflage, primrose, serendipity, sister, and twiffler.

I should add some of those to geirlyfr or something, but it's hard to present linguistic heritage discussions well. Lists of words that people can "ooh" over are okay, but perhaps there's a way to make it a bit more exciting...

by sbp on #sbp at 2003-11-05T23:09:39Z

PGP Snowflake Roundup

you are a beautiful and unique snowflake

you are a beautiful and unique fractal

oh my god! it's alive!

by AaronSw on #sbp at 2003-11-06T04:18:31Z

Idea: #sbp Review

Sometimes, you're on a channel where you need to keep up with what's been said on it (e.g. the name of the channel stupidly historically bears your initials), but you get on in the morning and there's a crapload of scrollback, or maybe you've been away for days and you really just want the highlights... The answer, often proposed in situations like these, is that some poor fool should keep a summary of the most interesting events of the channel and then provide an anthologic summary every week or so. Great.

But nobody ever does it.

Maybe it'd be nice to change this. I'd've liked a nice #swhack journal thing—the Weekly Swhack?--in'th olden days, but perhaps #sbp has enough cool stuff on it to merit one. Dunno.

by sbp on #sbp at 2003-11-06T04:32:35Z

Webic Variousity

I know it's old news now, but the illegal prime thing is still funny. As one /.er commented: "it should be noted that 31337 is prime, but its not illegal, yet".

In other news, there's the oou language (ta deltab), and the rather wonderful finnish phrases page: "What do Santa Claus, saunas, suicides, IRC, ammonium chloride, vowels with lots of cute dots on them, reindeer, Nokia cell phones, and gold-medalist ski jumpers who have recently turned into porn stars all have in common?"

But I rather like this story from an old /. post: "Alien arrives on Earth. Alien asks to view all Earth encyclopedias. Alien encodes all the content as a single very massive integer. Alien treats number as a fraction between 0 and 1. Alien takes out a crystal rod, measures, and makes a single mark on it. Alien goes home with the rod to decode later."

Good ol' miscellany.

by sbp on #sbp at 2003-11-06T04:38:58Z


Swhackit: it's like the Swhack weblog, but it isn't.

Links smorgasbords are becoming very popular, as they always have been in the form of bookmarks, but the information content isn't all that high unless they're things that you're noting for your own future reference, or unless they're heavily annotated with comments. So let's comment Swhackit! when possible, okay?

Out of the three content publishing from IRC routes being discussed (blog, linkblog, review) I'm leaning towards a review as being the most useful.

by sbp on #sbp at 2003-11-07T01:14:42Z

Idea for a Bookmarks System

One of the most annoying things about bookmarks is that generally you get asked what folder you want to put them into, and I don't like folders so I just put them into / and generally don't use them. Bookmarks for me are often just things that I need to be working with for a couple of days and then forget about. But given the linkblog format, I'm sure that actually I could make my personal bookmarks more useful.

I only have 17 bookmarks in Mozilla Firebird in linux at the moment.

Someone on #rdfig was musing about this a while ago, and decided they'd like a keywords based system. I think I'd like something like that: except I'd want the system to force me to type in a few keywords by which I could find it later... Perhaps epiphany has this feature, and perhaps it doesn't, but I'd also like to add quotes from the page and last modified time etc., so I'm thinking it might be a good idea to build a bookmarklet + Python server system for handling bookmarks.

The steps will be: you're on a page you like; select some text if you want to excerpt it... oh, this is annoying: people are giving me damn good information here on channel so I'm just going to ditch this entry for now and perhaps I'll write more when I've digested the #sbp info. Ta!

by sbp on #sbp at 2003-11-07T01:24:37Z

William Loughborough, He Be Blogging

William Loughborough has his own weblog now at his own site. William is... it's hard to say what William is and does and connotes. I think that only a stream of consciousness could come close: love, blindlessness, connectivity, integration, intelligence, eclecticity, musicality...

William introduced me to many notions, but also to an important organziation, the WAI, with whom he'd been working himself for some years. The WAI led me to many other things and people and ideas. He was also a co-founder of SWAG with Seth and Aaron and I.

I'm surprised that he hasn't written more publically available information, because he's such a stream of information that he's a delight to listen to and read no matter what he's working on.

by sbp on #sbp at 2003-11-08T00:44:02Z

Repetitive Editing

One of the things on my projects list is a replacement for sed. I've outlined many ideas for what I'd like in such a replacement, and I've been told that it sounds more like awk, but I think it may be somewhere in between.

sed and awk are, of course, well-used programs, heavily debugged and tested and proven on a large scale. Why would I go about writing a replacement? Mass reinvention of the wheel?

What sed and awk don't provide are, for example, scanf style routines for common datatypes such as URIs and datetimes. They don't do grapheme handling—in fact, how is their UTF-8 support? They're also highly dependent on the idea of field seperators—new lines for sed, and new lines and then record seperators for awk. What if I want to decr or recr a file? But most of all, I need something that can do repetetive editing jobs for me.

Repetetive editing jobs are those where, for example, you have to strip timestamps from logs, or reformat a list of googlecount results into a googlecount list query. It's any job where you're performing the same routine on a set of lines, basically. I'm sure that emacs supports that, and that sed and awk can probably be used for these purposes with a little investigation, but I don't think they were built specifically around my needs, and so I'd like to build something which is.

by sbp on #sbp at 2003-11-08T02:23:29Z

What To Do With Todo Items

In creating a more-optimal notes program, I'm currently thinking about the states of todo items. The creation of a todo item implies that there's a job to be done, perhaps by a certain date, so there should be an optional field for timespan. But the real world's more complex than that: many things don't have "do-by" dates, and those that do can often slide by varying amounts. Moreover, not all todo items get "done" to the degree of satisfaction that one would like. Structured procrastinators will also find that the mere existence of a todo list will prevent them from doing certain tasks. You'd also like to be able to say when you've done a todo item, and perhaps provide a link to the result. I could go on a lot further, but in summary, todo lists are a lot more complicated than they first appear.

Aaron suggests that having an AI program that asks you questions about your current state and then provides alternatives would help. It's a good idea, but the creation of such a program seems to be on his todo list...

by sbp on #sbp at 2003-11-09T08:26:56Z

The Keyword Problem

I like noets. Its only drawback is that I tend to post in #sbp, so people actually read what I'm saying. (Update: actually, X-Chat went barmy so I'm reposting this on #noets, so it'll probably be even more ramblier than usual.) It'll be better once Google picks up on it, of course. I search for scraps of things that I've noted via Google quite often.

Today, I'm wondering about keyword input. In my "b" notes program, a significant drawback was that you could only have one keyword per entry. Keywords are quite analogous to classes—in fact, in order to enable me to enhance the quality of the notes that I'm entering in my current "append to a file" notes program, I try to think of each keyword as a class.

This allows one to get around the keywords-as-phrases problem. Often, when prompted for keywords it's tempting to enter phrases; but that generally breaks the system, since one of the axioms of keywords is that they should be order-independent.

Anyway, this is one of the most banal and poorly-thought out of the noets so far, but I'm just trying to make sure that Googlebot comes across a fresh new page every day so that I can actually search through these things using Google.

by sbp on #noets at 2003-11-11T04:35:01Z

Bot Permissions

It'd be nice if robots.txt could exclude/include bot by purposes instead of names. What if I want to get rid of all permanent archivers, but let in search engine bots that cache on a temporary basis?

Then again, it's a wonder that robots.txt is adhered to at all.

by sbp on #sbp at 2003-11-11T05:29:25Z

The Clutter is Inherent in the Organism

Today's aphorism is provided courtesy of William Loughborough.

Here's part of my reply to him: "Part of the difficulty in capturing data is in capturing it as close to its original form as possible. Perhaps I've been reading too much Nelson, but I think that hierachies and lineralism are bad when they're *imposed* because they don't match working structures. So I've been investigating untyped links (Web), typed links (ENQUIRE, RDF), keywords (search engines), natural language queries (GNOME storage), and hyperintegration (Chandler, Edd's Desktop thing) as meta-structures. I've been investigating HTML, GUIs, SVG, JavaScript, ncurses, paper, IRC, and command lines as interfaces."

by sbp on #sbp at 2003-11-13T08:17:34Z

=> Wiki

Right. Since the previous post just about takes the biscuit for pointless entries, I think that a new kind of format is required. Currently, as you'll note from the title, I'm thinking about having blogbot append entries to various pages of a wiki. But I think that I'm also starting to get generally frustrated with the whole situation. I mean, once again I'm writing this as an IRC entry. Perhaps I ought to just make the logs available to the public. #swhack? Review? Same questions as always, same non-answers as always. Perhaps I ought to just start doing things for a change, actually trying things out; though I note that that's what I did with noets and it's turned out to be awfully misdirected. I really need some more server space—I've totally run out on, and is so slim that I have to keep it all down if I want to be able to use it for anything in the future. I'm starting to think I should give up most of what I've done already, preserve it as it is, and just start everything again on a new site. Though I think it's the Charlie-Brown's-new-life phenomenon—the strip where Chuck goes to Lucy the psychiatrist and says he wants to get away from it all, but Lucy says he'll just be the same old person as he always was, and to not even think about it, and furthermore that'll be five cents please. So on the other hand, it'd probably be best to avoid cluttering up a new site, especially since I'd have to pay for it. But a new site would enable me also to take myself off of Google. I'm not too impressed with putting everything I do in reach of Google without even thinking about it; I'd like to regulate what I do a little more, so that'll encourage me to work upon all of my non-technology related projects. Note that miscoranda (gah, this link format sucks—it's really hard to type) started off as something which was not generally technology inclined, but I noticed that I'd get more feedback when I did tech-related things. I suppose that people have come to expect RDF and Python hacks from me, and rants that can be used as Shirky fodder, but all that comes at the obvious cost of putting aside a lot of other interests. It's difficult to know upon which project one's time would best be spent, especially since most of the advantages and solutions that come from the projects that I work in are from the peripherally related benefits.

For example, on the Semantic Web, there's been very little of the core vision that's actualyl benefitted me directly. THe Simpsons data merging project that I did so long ago and the FOAFQ web form that I hacked up on top of what is basically libby's work have so far proven to be the only things that I recall have done anything for me. But at the same time, I've learned so much about recursive descent parsing, logic, data management, databases, knowledge representation, and so forth. And I've met many people, who've had many other exciting projects of their own. I've worked on the Semantic Web Agreement Group, for example, with William and Seth and Aaron—and that in itself was a project whose spin-offs were again more pragmatically inclined than the core goals. We didn't produce the fabled SWAG dictionary, really, but we got a lot of ideas out in the open, and learned a lot.

So I don't know. I've written a whole essay about this sort of thing already that I planned to publish to miscoranda but lost my nerve over. I was talking about the arts/science schism and how it shouldn't be affecting me but actually is, and my sites too. But it's all rather difficult to resolve since things tend to happen out of their own serendipity anyway, and the more you push the more resistance you get.

But this is another boring and banal post in something which has probably very little value even to myself in the future, so unless something very interesting happens in the noets area within the next week or so, I'll probably just take it offline.


by sbp on #sbp at 2003-11-13T08:30:52Z

Sean B. Palmer