I wouldn’t be so quick to assume the bulk of your load will be search. If 1 percent of searches result in an RSS subscription, you’ll quickly find that the automated polling of feed readers will outpace searches in network traffic. Perhaps you can define an analytics goal to see how many searches convert to a subscription. Maybe 1 percent is absurdly high or absurdly low.

On 3):
My feed reader supports hierarchic subscriptions via a folders mechanism. I can make a folder and place the artists I care about within it, or delete them should I stop caring.

Finally, you should look into an Amazon referral program. I guess Amazon has dozens of ways to handle referrals, so you might already be doing so. If not, think of it as a way to cover the cost of hosting and SSL certs Their referral rate is like 4 percent of completed sales, which ain’t shabby.

Back to forcing Apache to trust our shitty LDAP server.

1) This does need to be changed, I simply hadn’t thought too much about it yet. SSL would be nice, though I’m not sure it’s absolutely needed.
2) I’ve been wanting to do this, though it probably won’t reduce load since I believe most users will use the search either way
3) While currently there is no direct advertising, I could see it being in place later if the site gains a sizable audience. However, being able to create lists is important since I want one RSS feed with all the artists I care about and not individual feeds for each artist. I also don’t think RSS privacy is a problem, so that is not something I concerned about at this moment. I could see per artist rss feeds being useful, so I may add that in the future as well.

Thanks for the suggestions!

A few thoughts:
* Probably you want the admin interface on a less obvious url? And secured by SSL?
* Why not use slugs for URLs instead of numbers (ie /artist/stone-temple-pilots/)? It’s not like you handle multiple artists under a given query gracefully. That’ll reduce load.
* If the purpose isn’t to sell direct advertising to users, why not just generate RSS feeds per artist? No more email, no more passwords, and less server side processing to merge collections. This would also solve the user RSS privacy problem.

Anyhow, I thought links from the web, videos I like and the occasional status update from Identica would be interesting to some.

I wouldn’t recommend writing your own feed parser, it’s an exercise in insanity. Mark Pilgrim wrote a pretty good python universal feed parser. As an example of the insanity, consider javascript embedded in style tags.

My front page isn’t very interesting but I use the webspace for Mozilla Weave and Gallery2 (just installed last night).