Drunken Ramblings

This photo of the movements of a person around Paris for a year got me thinking about the narrowness of my own thoughts – how during the course of a week they go in certain patterns, around and around, and often thoughts lead to other thoughts, forming a path that sometimes I trace often. But as the months go by, new discoveries cause my thoughts to go in new directions, and every week I have a different set of thought-ruts.

I thought the concept of thought-ruts was cool, how the things we all think about over and over can tell a lot about us, and how much peoples’ ruts differ – the strangeness when people of completely different ruts meet, the conflicts when they differ in a violent manner, the goodness when they fit together well.

Then I thought it would be cool to map out thought – all thought, let’s say. At the bottom there’s the massive buzz of animals and insects, who’s brains just operate on simple rules that direct them towards food and mates. Above that there’s normal humans – the lower and middle classes, who are concerned with food and jobs and such. Above that are the artists and scientists and writers and stuff, who deal with abstract concepts like love and consciousness and identity and society. And above that there are the super-abstract thinkers like theoretical physicists, who think about the most basic nature of things, and mathematicians, who try to get at the basic nature of numbers which represent…. Too abstract for me.

So above the buzz would be the sparse few thought-traces of great thinkers, which arc up almost to … God? I don’t think so, but that’s the next step a lot of people would take…

I also read a book once about this person who could see people’s souls one day. They looked at a person and saw this magnificent, intricate light-structure floating in/above their head – with little bits of light going around representing thought and colors representing the emotions they’re feeling. Sort of like Dust in the His Dark Materials trilogy, where Mary Malone or whatever her name is can see dust swarm around people as they think – she looks at a child and it swarms around in fast little currents, and then she looks at its mother and the dust moves in more organized, purposeful patterns. It would be cool to see people’s minds as you walk down the hall – this person has a huge inferiority complex hanging on the side like a leech, this person has a huge self-esteem wall around a weak interior, one person says something to another person that punches a hole in their mind-structure, causing it to wobble and then start growing around the wound… Lots of possibilities.

This visualization of the blogosphere inspired me to think about visualizing web communities. People who know each other link to each other’s blog posts and call each other by nicknames. It’s a little closed circle of communication (you can see them scattered around the photo) in the middle of the Big Internet, which one of the community occasionally breaks by linking out to something the community usually doesn’t deal with.

It’s also interesting to look at how language is reduced as you journey towards the “epicenter” of such a community. An outsider, writing for an audience not familiar with a certain subject, would refer to to Ruby on Rails, for example, as “a web framework called Ruby on Rails”. As you read blogs of people writing for an audience more likely to know what Ruby on Rails is, they say something more like just “Ruby on Rails”, and then there’s the tight-knit group of contributors and such that refers to it simply as “Rails”.

Or take the creator of Rails, David Heinemeier Hansson. I wrote who he was because I figured people reading this wouldn’t know. Reading stuff normally read by people who do know, his name isn’t prefaced by who he is, and then at the epicenter of hardcore Rails fans he’s referred to as just “DHH”.

Extracting a social network from a bunch of web pages written by and for people who know each other is an interesting and daunting task. If you browse around for a while, you learn that this person created this, this person works for this company, they usually go by this username, etc. Important people like DHH are scattered all over the web – profiles, blogs, offhand mentions in hundreds of posts, etc. I once stumbled upon a blog by someone I sort of know from school – they linked to someone else’s blog, who I also sort of knew and they linked to someone else’s – I eventually figured out who they all were, but it was confusing because they were all going by strange little internet aliases and some had multiple abandoned sites and stuff… Reminds me of Google’s cool social graph API.

Well, I’m done now. That’s the problem with tumblelogging – blogging little bits of found interestingness like I’ve been doing lately, what chyrp is good for — every now and then you have to ramble like this and tie all those bits together. Ted Nelson was right – splitting things into little chunks is denying the basic structure of information – everything is deeply interwingled.

It’s cool how you can drag a box over an area in a photo in Flickr and write a little note about what’s in that area that people read by mousing over it — and in Facebook, how people put boxes over each other’s heads in photos, “tagging” that area as Bob and enabling links like “pictures of Bob” and “pictures of Bob and you”.

Too bad these tags, which add so much significance and meaning to images, are dependent on complex, site-specific javascript and server-side processes and database structure – when you move a photo from site to site, all that metadata is lost.

I wonder if there’s some sort of standardized, perhaps XML-based language or something for adding metadata overlays to images. I think I heard of this somewhere… Can’t remember

Here's a Cool Hack

I first saw it a while ago in in this rails ActiveRecord tutorial – as the guy plays with the object relational mapper in one terminal window, all the SQL it’s running comes out in another terminal window. I saw that he had typed tail -f logs/development.log in the second window, so I investigated: tail displays the “tail” end of a specified file; -f keeps it going, spitting out new changes as the file is modified. In the case of the video, the file was a log file that rails writes all the SQL it runs to.

So I did this with braindump. Now if you set log_queries to true in app/config.php, the system saves every SQL query it runs to a log file at core/query-log.txt. So now I can type tail -f query-log.txt in the terminal, and the next time I go to a page on braindump, it spits out the requested URL and then all of the SQL that was run in the course of creating that page, like so:

tail -f core/query-log.txt
# go to a page, and viola:
/pages/index
SELECT * FROM pages ORDER BY name DESC

It’s very useful for debugging, to see what’s going on behind the scenes. It’s also just cool to browse around and watch the terminal spit out SQL. It’s as if my program, braindump, is in the space between the browser window and the terminal window—you can see the cycle from HTTP request to SQL query and all the way back to HTML output.

It’s also cool to have the BQL console running in one terminal window and the SQL tail running in another. I type a BQL query into the console like so:

python xmlrpc_console.py
braindump> get creator of braindump

and immediately the other window spits out the SQL queries that my BQL parser needed to get the answer:

SELECT id FROM pages WHERE name = 'creator' 
SELECT id FROM pages WHERE name = 'braindump' 
SELECT object_id FROM triples WHERE predicate_id = 1 AND subject_id = 2 
SELECT name FROM pages WHERE id = 3 

and then the answer comes back to the first window:

python xmlrpc_console.py
braindump> get creator of braindump
pedro

Anyway, yeah. If you didn’t follow that, oh well. This is a show-off post, not an informative one :)

BQL is Braindump Query Language

I just implemented it last night. The idea is to create an english-like language á la SQL to simplify reading and writing information about entities in braindump. Right now there are four commands:

• set — 'set color of apple to red' => true
• get — 'get color of apple' => 'red'
• list — 'list where color is red' => ['apple']
• unset — 'unset apple' => true

It still needs some work, but that’s the gist of it. It’s mostly used internally now, but I also wrote a simple console program in python which lets you run BQL queries on a braindump server over XML-RPC. More about this later.

I just emailed Trevor about thinking chair and GMail gave me an ad for obese office chairs. (“up to 750lb. capacity!”)