Gallimaufry of Whits, by Sean B. Palmer

◇ Over the past couple of months I made a minecraft diary page, which is a backup of tumblr, using a tumblr archive script that I made. You have to download the data from the web first using wget, then the script extracts the posts from the raw data. The most interesting thing about the diary page is that it uses a Javascript tar parser to load the images. In other words, the hundred or so images are all in a tarball, and the script extracts them and loads them dynamically into the page replacing the static placeholder that it uses in advance. The script is called multifile.js and it's by Ilmari Heikkinen. With jQuery I only needed a few lines to get it working:

$(function () {
   MultiFile.stream('minecraft00.tar', function(f) {
      $('img[title=' + f.filename + ']').attr('src', f.toDataURL()); 
   });
});

◇ On 14 Jun we discussed buoyancy. We found that textbook explanations that say buoyancy is due to displacement are wrong, as the action of displacement has nothing to do with the buoyancy mechanism. Pressure gradient around an object is what causes buoyancy, and displacement is just a nice way of measuring the upward force against gravity generated by the gradient.

◇ I'd bought a domain called sbp.so on 9 May which was meant to be a frontend to my Dropbox folder for easy sharing. I decided though that this made the links look as though they were going to be persistent, which was certainly not my intention. At the moment I'm not serving anything from the site, though I am using it for the Tag URIs in the feed to this weblog. It's a shame because I rather liked the favicons that I used for the site:

[sbp.so favicons]

I'm also not entirely sure about the .so registrar, how long the domains will last and whether there will be any upheaval there. The name of the site was "Sean B. Palmer's Shared Objects", a pun based on the .so extension for shared object files on unix platforms.

◇ Also in May, I realised that I don't have a good intuitive understanding of the imaginary unit. Negative numbers I can get, in terms of direction, but for some reason the sideways direction of an imaginary number doesn't make intuitive sense to me. I did find out about gaussian primes though, which are prime numbers in the complex plane, and I fed a graph of them into a Conway's Life engine to make a pretty diagram. I wonder what primes are called in the quarternion space, and whether there are 3d cellular automata systems. The main problem I'm having with the imaginary unit, I think, is that I like to understand it in terms of geometry, and I don't know why you can have negative sides and positive area, but you can't really have negative area because you'd need imaginary sides.

◇ Back in March I read the Silmarillion. In terms of conception it may be the greatest story ever made by a single author, depending on the collaborativity of the Homeric works. The implementation is not as good as the idea, partly due to it being patchworked together from a heap of other works, but it's still very good. The original works are available too, and I bought a book called Arda Reconstructed which shows where each part of the work came from in the larger volmes called The History of Middle Earth. I've got all of the major Tolkien works now in the Harper Collins 2005 hardback editions, and they're very nice to read.

Before Judson Jerome attended a class on poetry at the University of Chicago, he wrote unrhymed free verse. His tutor was J. V. Cunningham, an epigrammist who distinguished himself from the prevalent fashion by writing in traditional, formal style. The class influenced Jerome so much that from then on he wrote predominantly metrical, rhyming poetry. A quote by Cunningham is used on the first page of a popular recent book about poetry, the Ode Less Travelled by Stephen Fry, broken into lines to make it look like an epigram:

[So far as I'm concerned,]
Poetry is metrical writing.
If it isn't that I don't know what it is.

The bracketed section is what was left out of the book. I could not disagree more that poetry is metrical writing. The Dictionary of Literary Biography says that Cunningham rejected “the usual Romantic and mystical pseudodefinitions” of poetry of the period. But I do not think that Cunningham's definition is any less of a pseudodefinition, and I will refute his before attempting to provide my own here.

Elton John was once asked to make an impromptu song out of something which was as prosaic as possible. An audience member had brought a vacuum cleaner instruction manual, and Elton produced a pretty good song from it. This was Elton making lyrical poetry out of the manual, without changing any of the words.

Now imagine if somebody made Hamlet: The Musical! All of the soliloquies would be sung as arias instead of acted. The effect this would have is also, sadly, easy to imagine. So dull things can be made to shine, and all that glitters is not necessarily gold in the musical West End.

In 1801, Coleridge translated a poem by Friedrich Leopold zu Stolberg-Stolberg, commonly known as Stolberg. He improved as well as translated it, saying that his version is “more poetical than the original”. Coleridge then gives a very literal translation of the original into English, without the punctuation of the original, given here with punctuation restored:

Still play, juggling Deceiver!
  Still play thy wanton Dances,
    Fugitive child of Vapor, that fervently temptest
      Onward the Wanderer's feet;

Then coyly fleest, at length beguilest into Ruin
  These maiden Wiles! I know them,
    Learnt them all, out of thy blue eyes,
      Fickle Nais!

And this is the version by Coleridge:

Lunatic Witch-fires! Ghosts of Light & Motion!
Fearless I see you weave your wanton Dances
Near me, far off me, You that tempt the Trav'ller
Onward & onward,

Wooing, retreating, till the Swamp beneath him
Groans!—And 'tis dark!—This Woman's Wile—I know it!
Learnt it from thee, from thy perfidious Glances,
Black-ey'd Rebecca!—

To check whether a poem is dependent on the metre, one good test is to make another version of it, using synonyms wherever possible and rearranging the clauses whilst keeping the sense. This can sometimes ruin connotations and other important things, so it doesn't always work, but as a rule of thumb it's a good thing to try. Here's an attempt at prosing Coleridge's version:

“You crazy sorceress-fires, you spirits of movement and brightness! I see you weave your gyrating dances by me now, then far away—you, who lure the wanderer ever on and on. You draw him to you, and then you recede, until the dark marsh below his feet mutters. I know this womanly way well, Rebecca, having learned it from you. I learned it from your deceitful looks, from your own dark eyes.”

This is quite a successful prosing, with a roughly similar amount of poetic feeling as Coleridge's version. In fact it loses some elements, some of the compactness and cadence which expresses the emotion of the talker, and gains some, such as the clarity of expression. Indeed, reading my version of the poem back helps me to understand the poem more. (Or poems, depending on whether you're a lumper or a splitter.) The prose version does lose a lot of the personality of Coleridge, but that is a problem which Coleridge would have been able to avoid if he had made a prose version.

So we found here that a little of the expression was in the metre, but that we were able to make up for it in another domain by making the poem clearer. So where is the poetry, and what is the poetry, if it isn't the metre? We could try to keep changing the words and the sense until it becomes non poetic, but we won't because we'd just run up against the Sorites paradox as people still can't handle radial categories and such.

Ezra Pound said that you can charge words using phanopoeia or melopoeia, using images and sound. I tried to remove the melopoeia as much as possible, replacing for example the alliteration of “weave your wanton” with “weave your gyrating”. But I also introduced some, in “womanly way well”, adding an extra “w” word compared to the original. It's hard to avoid. What is most prevalent in the poem in Pound's terms, however, is the phanopoeia, the images. The sorceress or the witch and her flames gives a mental picture very quickly, and it is developed very quickly into a great portrait of the irrlicht or ignis fatuus.

As well as the images, we have lots of metaphor. The irrlicht weaves its dances, and weaving and dancing are human activities. This makes us think of the irrlicht as possibly having other human qualities, of perhaps even being conscious. But the setting is a nocturne, this is a witch fire, so there is association with the dark arts, with the occult. These little metaphors are encapsulated in the great metaphor, or perhaps more strictly an analogy, that the irrlicht is actually the womanly wiles, the female way.

This larger analogy is realised in a strange way. What we start out with is a portrait of the irrlicht, sketched as an irrlicht, and there is no sign that it will turn out to be an analogy. After the extended description, we break instantly into the analogy: Stolberg just says “these maiden wiles” and Coleridge “this woman's wile”. We go straight into this surprising analogy, and have to realise quickly why it may be apposite.

We'll investigate one more example of the poetry from this Sapphic. The atmosphere of the piece is what might be called Romantic, in that it has gothic elements and subsists in the sublime rather than the beautiful. In fact I would say that when pondered carefully, the piece transcends the gothic and the sublime, making us think of a very charged and ethereal atmosphere, something that I call a metatrope. This is quite common in poetry, especially in the best poetry, but I'm not sure how much control the poet really has over this kind of expression, as opposed to sublimity, beauty, and the other features such as metaphor and phanopoeia.

But we have sufficient materials here, I think, to provide an answer. What is poetry? Metre, we have discovered, is not necessarily very important in poetry. When we looked for what might be important, we actually found many separate and interwoven things. We found phanopoeia, the projection of mental images and situations. We found metaphor, and that was used in two very different ways. And we found sublimity and perhaps metatropicality, charging the poem with an electric energy. There are probably many other things that we could find too, and all from a very short poem!

We find that poetry is made from a soup of important features, and that likely none of them are by themselves sufficient, and many may not even be necessary. Metre may count as one of these features too, but like the others it is neither sufficient, nor, I hope I have demonstrated, even necessary. The more of these features that some writing possesses, the more likely it is to be poetic.

A good closet play or dialogue could be written about a town watchman and a sexton comparing experiences from their nocturnal shifts. They would compare parapet to belfry, blazing torch to lanthorn, commerce way to process path, and mayor's pribble to vicar's prabble. I'd like to have learned about all of these things from an author of the early modern age, but of course they didn't write a manual about their period for someone from afar, with the possible exception of Robert Chester:

In yonder woodie groue and fertile plaine,
Remaines the Leopard and the watrie Badger,
The Bugle or wild Oxe doth there remaine,
The Onocentaure and the cruell Tyger,
The Dromidary and the princely Lion,
The Bore, the Elephant, and the poisnous Dragon.

The more famous Robert is a kind of reference of such things, of course. But the men of learning were pointed in their reading, and perhaps surpassed the genius of the age. The old sense of genius of course; there should be a dictionary that you can link to online when using old and strange senses. The tales should be coming from the house of Forman's vicar, or from the parsonage of James Newton of Old Cleeve, both of which I imagine would be of “humble roof”.

Do those guys have some kind of privileged view? There are many keen observers in any age, but they often get distracted with their own affairs. O for a chariot of fire to convey Borges to Dr. Donne, for example, so that the charge about divine keeping, that "even in him, in Christ Jesus himself, that admitted one exception; Judas the child of perdition was lost", could be met with the Tres versiones, and to hear the doctor's response!

I must say, I get almost personally offended when I find that my language doesn't have nice words for commerce way and process way, the roads to and from towns and churches and so on. Such words would be pregnant with connotations, of beautiful complex etymology, and be of unparalleled euphonic delicacy. Then again, we do have words like obeliscolychny for the light in the belfry or tower.

Even Milton from the light of his obeliscolychny would have been looking over the words of him who described Algarsife, so I suppose one can go too far with this. At least, it's nice to think that there's a point of connexion there, and such things teach us to be appreciative and keen observers, like the town watchman and the sexton are trained to be. The titular word, observicery, is borrowed and enlarged from one of the first works of an early modern playwright who wrote right up until the suppression of stage plays, James Shirley.

This is the story of how my server sploded, and what I did about it. Cody in his wisdom said that Ubuntu 10.04 LTS is “for wimps and dumb people”. The LTS version means that it will be supported by Canonical Ltd., the makers of Ubuntu, for many years. You'll have an old system, but you'll still be getting new packages and security updates.

Since this is for wimps and dumb people, however, I decided to upgrade to the newer non-LTS version, Ubuntu 11.04. To do that, you have to upgrade through Ubuntu 10.10 first because you can only upgrade in stages. To upgrade to 10.10, I simply followed the official upgrade instructions on the Ubuntu website. Everything was going fine until a screen about a program called grub, which loads up the computer, froze.

I exited the installation procedure and tried to start it again, hoping that it would pick up where it left off. It said it couldn't upgrade from 10.10 to 11.04 because something was locked, so that made it sound as though 10.10 had been installed correctly and that there was a file lock in place. So I rebooted. The system didn't come back up.

[Gandi Logo]

The server is hosted by gandi.net, a French internet services company. Their strapline is “no bullshit” and they're extremely popular. They happen to have an IRC channel on the network that I'm on, so I went in there after trying various things to get the server working again. The server was in an infinite boot loop which they couldn't fix, so they said I had to make a new server.

This was at 2am for me, so it was 3am in France. Thanks to flyou and mx, two extremely helpful Gandi staff, I was able to make an interim server just to put my main site back up for a while until I fixed it properly in the morning. In the morning I made a new server with Ubuntu 10.04 LTS despite having the option to use 10.10, and decided to serve all the important data and configuration files, except for SSH, out of dropbox. The server starts out with a 3GB disc, but it's on LVM so it's easy to resize. This is how I set the server up:

Resize the partition from 3GB to 16GB
Add self to sudo and disable root
Update aptitude packages
Install apache2, curl, emacs, irssi, and znc
Install dropbox
- Download the dropbox python script
- Link the computer
- Exclude non server paths in dropbox
- Make dropbox autostart
Configure SSH
- Update the SSHD config
- Copy keys to ~/.ssh
- Fix the permissions and restart SSHD
Link configuration files for emacs, phenny, and znc
Configure apache
- Link to sites-available in dropbox config
- Set ServerName to localhost
- Enable mod_rewrite
- Link to and enable each site in dropbox
- Restart apache
Link dropbox configuration binaries

The full instructions are available too. I call this the shoebuntu setup, because the server only has 256 MB so this is linux on a shoestring. It's been having Apache 2 memory problems, which I've hopefully managed to sort out using mpm configuration. The server is working well so far, and the dropbox syncing for websites and configuration is great.

An ancient Chinese philosopher asked in 300 BC whether it can be that a 白馬非馬, in other words that a white horse is not a horse. This problem is a trick of the light, a problem that can only exist because of language. Without the apparatus of language, nobody would think that a white horse is not a horse.

The fact that there are so many of these parlour games, philosophy that ought to be confined to party cracker jokes, has led to the ordinary language movement in philosophy. One can easily see why. Take a few of the propositions of Baruch de Spinoza, a famous 17th century philosopher:

In nature, there cannot be two or more substances of the same nature or attribute.
One substance cannot be produced by another substance.
It pertains to the nature of a substance to exist.
Every substance is necessarily infinite.
The more reality or being each thing has, the more attributes belong to it.

What Spinoza tries to do is to define words and concepts in a very limited and rigorous way, so that there can be no misunderstanding. He uses ordinary words but in new combinations, a bit like in mathematics. Consider how “complex number” has a very specific meaning which goes beyond the common concepts of complexity and numbers. Obviously the words chosen are apposite and point in the right direction, but one could not discern the existence of the Argand plane from the words “complex number”.

In mathematics this is not so repulsive, because of the unreasonable effectiveness described famously by Wigner in 1960. This means that maths tends to lead to further insights or even technology, more than we would expect from something purely theoretical. In philosophy, there is no unreasonble effectiveness, simply new constructs which are generally further abstracted away from usefulness.

The idea that philosophy should be a mock mimic of mathematics, with rigorous logical foundations, axioms and propositions, is slowly changing. The ordinary language movement did not get particularly far, but conceptual metaphor and related theories have helped things along a little.

Conceptual metaphors in particular are an extension of traditional metaphors, which in their extended form cover more kinds of thinking than most people realise. We are so used to using metaphors, in other words, that we stop realising that they are metaphors. “Things are looking up” is a metaphor, as there is nothing primarily better about upness. This is an example of the map–territory relation fallacy described by Korzybski.

Baruch de Spinoza found a metaphorical confusion when he refuted that God has certain attributes of a person. He notes that after people “considered things as means, they could not believe that the things had made themselves; but from the means they were accustomed to prepare for themselves, they had to infer that there was a ruler, or a number of rulers of nature, endowed with human freedom, who had taken care of all things for them, and made all things for their use.” He holds this to be a fallacy perpetuated by the simple metaphor, when the prejudice turned to a superstition.

The very idea that philosophy should be mathematical is a metaphorical confusion. There is nothing inherent to the pursuit of philosophy that dictates it should be formed in terms of mathematical style logical enquiry. There is no reason, for example, why advances in sociology should not produce the best philosophical findings. There seems to be no particular reason why art or politics should not be able to produce the same. Stranger insights than simply switching the domain of discourse metaphor may be more fruitful, but only if we are receptive to unusual directions.

The scripting languages we use as of 2011 are well over a decade old. Perl, 1987. Bash, 1989. Python, 1991. Lua, 1993. JavaScript, PHP, and Ruby, 1995. They have undergone many revisions and have many libraries and techniques, which is how they are managing to stay usable and relevant, but clearly there is no popular modern scripting language. Only Perl6 shows any prospects of becoming one.

People do try to create new programming languages from time to time, but they fail to achieve momentum. This is because they have to compete with these established languages with their many modules, packages, and libraries, and tutorials and developed techniques and shared experience. To become popular, a programming language would have to compete on this level more than on the innovation level.

Yet language designers are often looking for new features that will make their language better than others. They do not instead think that a good new language could be most remarkable for being entirely unremarkable. Instead of pushing the edge of programming languages, why not clean up the core? Beauty can come from eliminating the burrs which are so prevalent in existing languages. In short, such a language would have features like these:

Clear and unobtrusive syntax
Small, learnable grammar
Ability to use many paradigms
Proper unicode support
Rich mathematical expression
Speed, especially of initialisation
Well tested, portable code
Large standard library
Modern package management

It is not worth debating individual choices. We may naturally wonder whether these points mean, for example, that indented code is more or less obtrusive syntax than braced code. These choices are what the designer has to make, and they have to make them work. They are based on intuition and experience. What is important here, on the other hand, is how these features relate to the overall characteristics of being unremarkable and well balanced.

Being well balanced means that a language must eventually manifest all of these features in equal concentration to be successful. If the syntax is painstakingly developed but there are no libraries, that is almost as bad as a terrible syntax with libraries—but at least libraries can be made later. They must end up being worked on in equal concentration, but they don't have to start out that way. The earlier things on the list are those which are most important to get right in earlier order. But each one is equally important in overall terms.

This method means that there is very little room for creativity in making a modern scripting language. Most of the work would be in terms of doing huge, expansive surveys of existing languages and matching their core expressivity; of surveying many programmers to make sure that the syntax is huffmanised against their practices; and of making sure that the unicode, maths, and package systems are watertight and bulletproof.

Most lightweight markup languages such as Markdown and Textile invent their own syntax for links. In other words, they don't use similar syntax to one another, and there is no particular internally consistent logic to the syntaxes that they choose:

Markdown	This is an [example link][http://example.org/].
reStructuredText	This is an `example link <http://example.org/>`_.
Textile	This is an "example link":http://example.org/.

When converted to hypertext and rendered in a browser, links are typically underlined and coloured, with the default colour being blue. Since this is a kind of emphasis, why don't these languages use their default emphasis as the core of their link syntax? Such a syntax may look like the following, imagining that *asterisks* are used for emphasis in this language:

The *commonly used* http://example.org/ example domains *changed recently*,
tech.slashdot.org/story/11/01/30/1823241/Examplecom-Has-Changed and now
redirect to an *IANA page* www.iana.org/domains/example/ instead. The IANA is
operated by ICANN.

*[ICANN Logo]* inamidst.com/stuff/2011/icann

*The ICANN Logo*

Which would be rendered like this:

The commonly used example domains changed recently, and now redirect to an IANA page instead. The IANA is operated by ICANN.

The ICANN Logo

This can easily be done using a small python script, so the syntax is programmatically lightweight. In this example, strimming the protocol from HTTP links is allowed, though this does make it difficult to use local links. It would depend how often local links are used as to whether the protocol strimming should be allowed. These examples have all been centred around inline links, but of course this technique would work for reference links too:

This is an *example link*. [1]

[1] http://example.org/

Image links have been indicated by having the link text be interpolated using square brackets. This might not be an optimal syntax either, especially since for example specifications sometimes use [TAGS] in brackets as link texts to citations in reference sections.

The python script doesn't support lists or anything like that, but it does support literal use of asterisks in text through \*slash escaping\*. The intention was to use this for jotting bits of hypertext, but HTML is okay even for quick jotting in most cases. So this may be useful instead for incorporating as a technique into new more fully featured lightweight markup languages that want to have a consistent and reasonable design.

Back in April, I found that I had hosted one of my earliest pages on the web at five separate free web hosts, and used two link shortening services, to make the page unlikely to disappear from the web. Over ten years later, not a single one of those services is still available, even the link shorteners, and not a single one of the pages was stored in the Web Archive or any other cache that I could find.

The blog that I did most of my writing to between 2002 and 2006 isn't on the web anymore, and I wanted to point a friend to one of the files recently. I made a Single Page Application out of it so that I could point to it more easily, but I'm still not hosting it in a particularly stable way. It doesn't really matter too much because I don't link to most of that old data anyway, though it is interesting to read through it.

Not many people link to the things that I create, perhaps in part because I don't keep them in a very stable way, but also increasingly because I'm unwilling to put things where I'm going to move them anyway. I'll email things to friends, or use pastebins, or use secret addresses. If somebody wants a reference to it again, they can just ask me.

Similarly, now that I'm writing these particular comments on my own domain, hosted on my own server, I'm not inclined to make permalinks. This would make it easier for people to link to, which I'd like, but it would make me feel responsible for the links, which I don't like. I want to move stuff around, to reorganise constantly. I don't like keeping things in one place, which the web requires for referential stability. So ultimately I don't really like publishing to the web because of this.

The idea behind the Single Page Application was to try to make it so that if a particular link inside a system breaks, then it has a kind of sprung fallback, something that I've written about at least partly before, though I thought I had a more specific piece about that too. Post 41 in the 2002–2006 blog was about a similar kind of idea too, using fragment redirects to provide catchment areas. But I don't think the idea is viable, because you still have to maintain the fallback or catchment area, even if the inner references can be more fragile.

I had been thinking about manually putting links in the bottom of my entries in case I wanted to make something linkable, but I'm not entirely sure where I'd put the stuff anyway. Sometimes I think that packing a lot of things together in a .tar.gz archive and throwing that on the web with a short description would be best, for example with the Voynich manuscript experiments I did recently, which contains tons of disorganised scripts and notes. In general, I think the web just isn't suited to this kind of data and management.