Time Heals All Wounds.. And Then Kills the Patient
<Previous Next>
Mon Sep 11 21:09:01 2006
Sharpening one's Metaphors

I'm working on improving the LJ-POUND linkage (right now, POUND is relatively dumb and will re-post entries to LJ if I update them on my software). I have determined that while posting to LJ is easy, and blind batch manipulation of LJ entries is easy, finding a particular LJ entry programmatically isn't quite that simple. LJ does a shoddy job of indexing its posts - the best pseudoindex one can get is to ask for a particular date, get a list of entries on that date, and then plod through each of those trying to recognise the entry there. Entries do have a timefield, but LJ's concept of dates leaves a bit to be desired -- they do something funny with timezones (presumably), so your entry may be an hour off where you expect it to, programmatically (I say may be because looking at a set of entries, I see no pattern in when they're an hour off and when they're not). Oh, and LJ doesn't keep track of seconds, so you get to chop seconds off if you're doing a comparison. This up-and-down truncation means that if you want to robustly look for an entry, you might get to look in three spots, one for the date you wrote it, one for the date before (in case you posted within the 59 seconds after midnight and want to be really sure), and and one for the day after (in case it decided to add an hour to your after-23:00 post). You also get to try adding an hour and subtracting one in your comparison to deal with the random adding/subtracting hour thing, and if you should so happen to do two posts separated by exactly an hour, you simply lose if you want to deal with them without trying to recognise content. I'm going to sit on this problem for awhile to see if I can figure out more of what's going on. There *is* something called an ItemID that comes out of PostEntry() that maybe I should be using instead -- I could save these in the database whenever an entry is posted to LJ and maybe avoid this stupid mess.

Quirky, interesting things:

On Wikipedia, I sometimes *shocking!* find myself disagreeing with the way the community or even Jimbo go on some issues. Mistakes and unfortunate things happen, and I find it tough to tell people that I think they were wrong on some issues -- sometimes I do, sometimes I don't. A certain amount of social capital is spent everytime one does so, and things by and large work more smoothly when one is quiet, but it's not particularly healthy when people never see other people worried by actions they do that might be questionable. One recent example involved some girl (a fairly new Wikipedian) uploading pictures of herself to illustrate some of the sex-related articles on Wikipedia -- there was some discussion on this, in particular if it was really her in the pictures, along with the dishonest maneuvering of both prudish factions (who would prefer such articles not exist and if they must, that they not be illustrated) and .. factions of people I'll call "pigs" who were cheering her on just because they like tittilation and presumably don't know about 4chan (or more realistically, like personally cheering on erotic "stars"). The argument was predictable, apart from the dishonest factions who did their best to find every technicality to keep or delete the images (a stupid, lawyerlike mentality), there was the usual debate between people who think an encyclopedia should be family-safe and careful, and those who are aiming for an encyclopedia in a more pure, audience-ignoring sense. I'm among the latter, being largely social libertine - I don't think Wikipedia should try to be safe-for-work or for children. Just as we decide it's more important to show the Danish cartoons on Mohammad in the article on the controversy, which we knew would offend many, many people, we should decide to accept pictures that help illustrate articles, even if they offend or arouse. There was, in this situation, some question on if the girl was uploading pictures that were really of herself or if she was simply taking images from a paysite and removing attribution. Alternatively, there was the concern that she was of an age sufficient to make it legal to share said images. Instead of carefully and delicately asking these questions, Jimbo decided to jump in, suggest she was trolling, and she was blocked almost immediately afterwards. I think this is unfortunate and bad judgement on Jimbo's part. The images reappeared on the horrible-but-sometimes-amusing Wikitruth site, one of the many sites devoted to criticising Wikipedia. Despite some occasional actions that I consider unfortunate on the project, by-and-large governance of Wikipedia is done well - the sites devoted to criticism tend to be run by mentally unbalanced folk with unintelligent criticism. That doesn't mean that I think that the right decisions are always made by my support of and involvement in the project.. On the other side, there's the problem that people enforcing policy to the extent it needs to be enforced (especially policies that are bound to anger when enforced, like reminding people that user pages are not personal webpages or making sure we don't run afoul of the law) tend to become unpopular, and our notion of consensus resembling, to a certain extent, voting, people doing what is necessary and good tend to harm their ability to take positions of responsibility. It's a difficulty in any electoral system when the demos don't understand or take responsibility for projects/societies. Apart from better ways to manage sufferage and/or inculturation, I don't know what the best way to manage the impact of these problems is.

I think I mentioned Malkit Singh before -- an indian pop-Bhangra musician who I saw on TV at India Garden before -- his clever music video for Jago Aaya helped me notice the song, which is an excellent song to program to. Having found more of his music, I've decided that I really like his sound - like many people who get a programmer's high, I have my own preference for music that helps me get in the zone. Almost none of the stuff has recognisable lyrics, and when it does, it's generally not stuff that grabs too much attention -- I suspect that music with lyrics tends to interfere with the word and spatial manipulations involved in weaving code. I also suspect that the music I'm listening to tends to influence my pace and style of programming (although I have no way to know for sure because my human introspection is turned off while I code). For me, Bhangra is great for coding - the constant rolling beats help keep me moving and inspired. On the occasions I need to stop to think, I typically switch to darkwave (not quite ideal because it has words, but Darkwave sans words becomes Trance, which I've never liked, either for programming or for casual listening). To my fellow readers who have programmed enough to enter Deep Hack Mode, what kinds of music help get you there? Do you vary the music for particular things, and what for?