All 2 entries tagged Rant

short attention span leads to madness….

Yes, once again I've got bored waiting for a football match to finish and have wandered off to write blogs entries while most of the population sits glued to their seats. I can't help but think that the sole purpose of extra time is to irritate people like me who have just about learnt to sit still through 90 minute games.

But, I'm determined to pretend that this is at least a pseudo intellectual post, and so I'll start the following rant: what's most irritating about football is not the game itself, but the (oh, that was a goal, I can hear the shouts from next door) encyclopedic knowledge of certain fans (mentioning no names….).

While I confess that it hardly gets my pulse racing, I'm perfectly capable of watching a game whilst enjoying a pint or two, and while I really don't give a toss who wins, I can just about recognise a particularly skillfull play and get a dim sense of recognition of the level of talent involved. I begin to feel the hounds of death pursuing me when the conversation turns from the game at hand, where I can at least make naive observations and ask stupid questions, to discussions of the players' histories.

Before long, I begin to suffocate under the weight of footballing trivia diffusing into the atmosphere, and soon the game at hand has faded into insignificance in the face of potential questions concerning the crucial period during the last decade in the history of a particular (oh look, an interruption: apparently Argentina scored that goal just now. It was a great goal. Oh and I am informed that play is apparently not a noun one is allowed to use to describe football, at least in England.)

…

Where was I? If Isaac Newton had had to put up with such interuptions, we'd never have discovered gravity.

…

The point is that game that is apparently the focus of all this activity, or even the sport itself is of course of negligible significance. What is actually important is the act of following the sport. This is why it is of far greater significance to ascertain from one's fellow sport fan not what their opinion might be on a particular aspect, say, of a game, but instead on their opinion with regards to a an aspect of what one might term the metagame, viz what team they choose to follow. Perhaps I was wrong in my earlier rant with regards to football, perhaps football represents not primitive tribalism manifested in a modern age, but instead is essentially an insecurity of identity.

In other words, we (well, not me personally, of course, lesser beings, amongst which I shall assume for the sake of rhetoric I fall) construct for ourselves a metagame, a semiotic universe of sport in which obscure trivia are elevated to heights of absurd significance, purely for the purpose of using it as a tool of social classification. We are thus able, through the application of a few judicious questions, quickly able to situate a new acquaintance in this metagame, and are thus able to establish not only our derived position respective to them, but are hence able to establish a mutual identity.

(This ties in with my sudden conviction that all identity is inherently subjective; formed only from interaction with others: before he met Friday, Robinson Crusoe did, in a sense, not exist.)

As a proof, or rather as a potential line of supporting evidence, consider the odium with which are considered pretenders to the throne of fans: those foolish individuals who claim a greater knowledge of a sport than they in fact possess. Is this not a demonstration of a fundamental insecurity in the heart of the accusers? Is it not the case that –

Bugger. I've run out of beer.

: 24 Jun 2006 22:34 |  Tags: Football Rant Sociology Sport |  Comments (4) |  Close comments |  Report a problem

What's wrong with computer interfaces: a rant

One of the greatest advantages of computers is that they can easily allow you to look at things in different ways. The problem is that they rarely do so.

Think about files in windows. Traditionally, each discrete "file" existed (in a non–real sense) in a discrete location, a specific, and specified, file directory. (More recently renamed a "folder", persumably in a spirit of user friendliness.) In order to find a file, you had to know its location, or perform a cumbersome virtual "search", that, like a real search, scanned directories one by one. This is still the model used today, in Windows XP. It's easy to see that this model parallels physical organisation, books on shelves (with each book on one and only one shelf), and so on. I shall call it the physical model. It functions perfectly well provided that the artifacts in question can be easily divided into categories, without overlap, and provided that the user is either well organised or has few artifacts to organise. This is rapidly ceasing to be the case with the modern computer user.

Two areas in which shortcomings are becoming particularly apparent are digital photos, and digital music files. For these files, there exists an alternative model of oganisation.

(It's interesting to speculate which areas might be next to undergow massive growth. Digital photos have multiplied with the proliferation of cheap, "good–enough–to–make–prints" digital cameras, and digital music has been spread as a result of: (in chronological order) easy (illegal) music downloads at "good–enough" quality concurrent with easy cd ripping, cheap (and fashionable) digital music players, and finally easy legal music downloads. Although video files can also be spread through similar media, current problems exist with: the slow speed of downloading large files, even on high speed broadband connections; the lack of a "good–enough" standard for files in terms of quality (it needs to be to DVDs what MP3s are to CDs. Divx files, in my opinion, aren't), and teething problems with portable players. Ebooks have been technically feasible for years, but fail to inspire much demand.)

I'm going to speak mainly about music files, as they nicely illustrate the point I'm making.

Now, when I started to accumulate MP3s, I dumped them all in one directory, loaded them into one playlist in winamp, and played them randomly. Now I've thousands of files, stuck in various directories, and use four seperate programs in an attempt to "manage" my collection – Winamp, Helium Music Manager, Media Monkey, and Tag & Rename. All of these programs use, in various ways, a database. The database model involves, more or less, assiging data tags to each file, indexing these data tags, and perfoming queries on these indexes. So music files are tagged with data fields containing artist name, track name, album name, release year, and so on. It is then possible to search for all songs released in 1967 by the Beatles, and so on and so forth. The programs I use all manipulate these databases in various ways.

Tag & Rename, I use for actually writing the data tags, as it's fast and easy to use; and Winamp I use to search for files, as it's extremely fast, and to actually play them, as it's possible to add various plugins to improve playback quality. Helium could, in theory, perform all these tasks, but it's slow and cumbersome to use. (Media Monkey I use solely to synchronise files with my portable player, something Helium doesn't do.) Helium is, however, extremely powerful. There's not much you could do to improve it, in terms of database–based capabilities. So it nicely demonstrates both the advantages of the database model over the physical model of file storage, and the shortcomings of the database model against what should be possible.

Firstly: the various programs use slightly different file tags. Now, this is partly my fault for using different programs, but given that (for example) there exists a standard tag field for a rating, you could reasonably expect that all programs that access it to treat it in the same way. They don't. Winamp ignores this field, and stores ratings in the internal database, meaning that a) ratings are unaccessible to other programs, and b) if the database is lost (as it is, for example, on a reinstall) all ratings are lost. It's possible to use an external plugin (which neads to be manually applied to each file after every change), in which case Winamp marks ratings 1 to 5. Media Monkey uses the standard tag field, rating files 0, "bombed", 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5. Helium uses exactly the same field, but omits the "bombed" rating. This means that a file rated 1.5 in media monkey shows as rated 2 in Helium, and so on. I don't quite understand what Tag & Rename does, but it definitely doesn't do quite the same as the others. Not all programs use the same tags, with Helium having at least twice as many fields as Winamp. It allows tagging with data for "mood", "situation", "release type", and a great deal more. Winamp doesn't. The point of all this? The need for integrating of all file tag fields into a higher level, so that all programs access them in the same way. This will be important later in the discussion. (It is more or less the case with digital photos, as far as I can tell, that contain tag fields for shutter speed, aperture, and so on.)

It's important to realise just how much data can be added to files. In Helium, fields exist for: Artist, Secondary Artist, Artist Sort Order (to ensure that tracks by "The Beatles" are sorted under "B"), Title, Subtitle, Remix, Title Sort Order, Genre, Mood, Situation, Language, Comment, Tempo, BPN, Recording Year (of track), Track Number, Total Tracks, Rating, Preference, Quality, Studio, Composer, Band / Orchestra, Conductor, Lyricist, Lyrics, Album, Album Subtitle, Album Sort Order, Series, CD number, total number of CDs, Release Year (of CD), Media, Catalog Number, Produced, Published, Copyright, Grouping, Musician Credits, Involved People, Arist Biography, Artist URL, File URL, Buy CD URL, Publisher URL, Radio URL, Encoded by, Software setting, File Owner, Picture, More Comments, Orginal Artist, Original Title, Original Year, Original Lyricist, Original Filename, and ISRC. Obviously, not every file need have every field filled, but this illustrates the sheer volume of data handled by these programs, and, presumably, explains the slow speed of Helium. I should add that it's possible to download much of this data from various internet servers, though obviously quality varies.

Having entered all this data, it's then possible to get Helium to reproduce it in various permutations. This is actually better than it sounds, as amongst the things possible, one can browse by album, search the various fields, surf through a web page–like interface, call up auto searches for, say, all upbeat songs produced in the 90s that haven't been played in the last month. Or, indeed, all 5:00 minutes songs with "t"in the title suitable for a party. It really is an extremely powerful program that, once one's mastered the somewhat cumbersome interface, allows for extremely sophisticated data manipulation. So what's my problem?

Basically I'd like to question how appropriate the database model is to personal files. The fundamental ideas of the database system are these: the physical (or pseudo–physical) location of the artifact is irrelevant; the metadata is entered initally, a single time only, correctly, then never changed except for minor updates (file and forget, one might say); and that manipulation takes the form of extended queries on a higher level. Problems exist with each point.

Firstly, and least importantly, it's still sometimes necessary to know "where" files are. Backups, for example, are always easier if one can backup only the files needed, and usually this comes down to specifying a location. But this would be easy to overcome.

Secondly, metadata is not unchanging. Particularly when one might have thousands of files with incomplete information, it should not be necessary to input massive amounts of data. When data fields take the form of "ratings" and the like, one would naturally wish to move files from category to category as tastes change. This should be easy to do, but I find it strangely difficult once the database has been constructed. I have a directory of about 700 "odd" songs not from specific albums (but instead from "best of the 60s", Now 54, and such like). If I notice that, say, one of the songs has the wrong artist. If I try in helium to change these details, it goes through referencing each and every one of 700+ songs. It's a bloody nightmare, to be frank.

Thirdly, and fundamentally, I would question the single level of data analysis. This is where my suggested model comes in. Only in certain circumstances do we best analyse things by lumping them together then extracting parts of them. What we really do is divide things into groups and then various subgroups, depending on need. The advantage of virtuality is that the same things can be divided many times simultaneously, allowing, say, the picture of you in the garden to show up in the group of "you–at–home", in the organisational area "me in various places", the group of "casual–photos–of–staff–relaxing" in the area "our team" in "corporate photos", "me–improving–the–garden" in "my projects", and so on and so forth. Likewise the same music track shows up in "party", "favourites", "old memories", and so on. I shall call this the metagroup model. It is possible to mistake it as identical to the database model, but the truth is that the database model can be made to function as a poor imitation of the metagroup model. To illustrate:

Helium has a tag field for "genre". Let's see about organising our music. The categories are ridiculous – having accumulated much music from various sources (all legal, I hasten to add) each with its own system, there isn't much overlap. The first few categories are: 70s, 70's (note the different spelling), 80's, acid jazz, acid punk, acoustic, acoustic rock, alt rock, alternative, alternative and punk, alternative rock, alternrock, ambient alternative, art rock, and avantgarde. Then on into b, c, and down to vocal, wave, and world.

First problem: "70's, 80's" vs "Acid Jazz". Obviously these are two levels of analysis. (One might question the extent to which "80's" is a useful category, given that it's possible to search by release year, but there does exist music that's "quintessentially 80's" i.e. the stuff on "best of the 80s cds", and this doesn't include everything produced from 1980 to 1990.) Furthermore: "acoustic" vs "acoustic rock" vs "alt rock" vs "alternative" vs "rock". Now, to me, "alt rock" is the zone of overlap between "rock" and "alternative", while "acoustic rock" is "rock" with the additional property of being acoustic. Is it possible to have "Indie" anything, or only rock? Can "Pop" be acoustic? And so on. Now, I don't mean to deny that it is possible to come up with a system of tags that answers all these questions, but I would argue that any system would be extremely idiosyncratic. It would also have to be constructed in one fell swoop, as it's extremely difficult (and tedious) to rearrange tags.

What a true metagroup model would do is allow visualisation and manipulation of multiple hierarchical groupings. So all files existing in a "physical" level could be added to or removed from multiple "virtual" levels that could themselves be divided into groups and subgroups. Hence one level might be genres, another situation, and so on. In fact some of the tag categories are quite useful, but others aren't, and the main problem is that I can always come up with plenty of possible categorisations that aren't covered in the tags. So can we all: it's personal to each user. Say I want to consider "World Music", I have to blanket tag all the relevant files say genre=World Music (which of course precludes say Latin American pop from showing up in the 'pop' category). What I want to be able to do is create a new level of analysis, with "World Music" versus everything else. Then I might want to divide 'classics' from stuff that's a bit dated from stuff from my childhood from cutting edge music. A bit later on I might want to divide my music into, say, stuff my girlfriend likes versus stuff she fails to appreciate. Then later on I might want music for parties versus music to play in the car. So, should my girlfriend plan a party for middle aged international diplomats, it would be child's play to create a playlist for the occasion.

(A (very) significant by–product of my multi–level approach would be that the computer could create a virtual level and manipulate that: in effect it would just hold lots of lists and move files about. This would mean that it each list entry would just be a shortcut to the file itself, and so each re–organisation would just mean creating a list, without having to write to the files themselves. This would mean a far from negligable speed increase.)

Well, I hear you say – that's your problem for not organising your music properly, and indeed I'm probably exagerating the amount of grief it actually causes me. But it is awkward, and really indicates the faults with the current computing paradigm, which can only get worse: Windows Vista will utilise a database system of indexing. (I think. I may be wrong, as to be perfectly honest it's extremely difficult to extract useful information from all the publicity crap they put out…) I sincerely hope this won't be the case, but I expect to have to wrestle with this level of irritation for every file I wish to sort. I'm not quite sure how the organisation will work in practical terms, but I expect that Microsoft's idea of useful categorisations will be as relevant to my computer usages as is, say, that bloody paperclip.

I wish I could design my own shell for windows, but unfortunately, I can't. Nor, really, can I use an off the shelf program. Interestingly, when I visited a friend of mine the other day, he'd found some obscure DOS program (Lotus Agenda, to be precise) that would do nearly everything I wanted to, albeit in a somewhat awkward fashion. What it wouldn't do is offer shell integration: I'd have to create a database by hand, and update it by hand everytime I found a new file. Essentially I'm not going to get what I want until computer program designers stop trying to predict what users might want to do, and actually ask us. Or ask me at least, as most people are probably quite happy using the default options for everything. Or, more acurately, they probably never bothered to think that things could be any other way. We are in effect trapped into a paradigm that was invented for us, and most of us are quite oblivious to the assumptions that went into it. So what's the solution? God knows….

[Wow. I got a bit carried away here, didn't I? I haven't quite got over how blogs allow you to just rant about random stuff….]

: 16 Jun 2006 18:48 |  Tags: Computers Rant |  Comments (4) |  Close comments |  Report a problem