Oh how ignorance was so blissful. You never realize it until you learn something new ...

I have recently added TONS of new bloggers to my daily "to read" list ... a lot of technology, venture capital, designer blogs, etc.... and it's all making me realize the folly in doing things wrong.

It's so hard to do ONE thing right ... and it's about 1000x harder to do tons of different things right. Perhaps this is why web development on all my projects are so stalled; I'm so intent on doing everything perfectly that I rarely get stuff done.

It's quite hard to not only plan the website, but to create the website with solid scripting ... backed up with a good design that is also standards-compliant and semantically correct, while making sure the UI flows well and the iconography isn't weird ... it's really hard.

A new read I added today is Widgetopia, which covers a variety of subjects which all deal with how users interact with websites to make the work. And reading over a lot of the pages ... I just want to revamp a lot of my websites from scratch (again).

Of course, reading Garrett's entry on cleaning up his website makes me want to take the time and actually execute my plan to clean up the Tabulas system ... AHH TORN BETWEEN PROGRAMMING AND DESIGNING. SOMEONE HELP ME.

In any case, I've been working on Audiomatch lately. Neeraj rewrote the plugins (so they're a lot better, and now we have two-way communication between the plugin and the site), and I've been revamping the design/backend of the site so the display is better.

We're taking the site pay, but we're also trying to make it better. Besides the changes I've implemented on the design side (semantically marked up HTML, CSS-driven, better looking, etc.), I've also begun to set a wider range of "community" features to Audiomatch.

One of the cool little features we can actually afford to run are customized reports for each user at the end of each month; this report will contain the songs they've listened to ... but it will also allow me to run a program that will "suggest" artists for them based on the listening habits of other users. On top of this, I want to see if either Neeraj or I can build some sort of comparison engine for users and specific artists; e.g. which users listened to which artists most frequently; we can also suggest "friends" to add to your list.

The idea is to take Audiomatch beyond it's simple beginnings to an inherent music suggestion engine; based on the listening habits of our 15K+ users, we can figure out what artists are similar (it already works to a pretty good degree on limited data!) ... and then make suggestions to those users.

I know I'm always looking for some new music, and it'd be so nice to just have a tool that could suggest artists for me.

I've also been planning the resurrection of the lyrics database; I have about 60,000 lyrics at my disposal right now just stored in the database, ready to go. I'm thinking of building a wiki-like PHP lyrics system; allowing multiple categories of comments per song (what did it mean to you, song facts, song meanings, general comments). I can use some sort of user authentication/karma system to degrade bad comments over time so the system will self-moderate.

As for the image generation, I'm getting quite frustrated with how to handle. I just don't think ImageMagick is going to scale to 15,000 users very well... I may simply stick to the built-in GD libraries ... allow people to upload "background images" for their Audiomatch and have them figure out what the X-Y coordinates should be for the text.

But then I also have to decide which fonts to use. I tried to install a TTF font reader, but for whatever reason, the fonts come out broken when I generate an image with them. I'm not entirely sure why.

Yeah. There's been a lot going on in terms of development.