The perfect aggregator

I've been thinking about feed aggregator software, and playing with Bloglines. I'm a SharpReader user, and before that I used Syndirella, but the more I think about it, the more a web-based, serverside aggregator makes a lot more sense to me. I don't need an aggregator that embeds a web browser. I already have a web browser which is always open, and the fact that SharpReader embeds IE while I use Firefox just causes me grief (if popups are irritating normally, they're doubly annoying in an RSS aggregator -- and they seem to make SharpReader unresponsive for several seconds at a time).

Bloglines is a free service that offers a web-based aggregator. It came to me highly recommended, but the more I used it, the more I disliked it. Oh, the interface was polished on the surface, but it didn't seem to do anything I expected an aggregator to do. I added MetaFilter to my list of feeds -- it showed 3238 unread entries. I clicked on the feed name and Bloglines proceeded to load all 3238 entries in the right hand frame -- a 1MB HTML file that brought Firefox to its knees. Then I realised I couldn't mark individual items as read, or delete them. No, it was all 3238 or nothing (and they'd all been automatically marked as read when I opened the feed, anyway).

Okay, so the size of the page is a relatively minor UI complaint, and I don't actually want to use Bloglines to read the Metafilter archives. But that situation is different only in scale to one where I was reading 8 new entries from somebody's weblog or the New York Times, but didn't want to read them all in one session. That's not unusual behaviour. And what if I wanted to keep an entry around in some fashion for later reference? Bloglines fared a little better in this respect, allowing me to add entries to a 'clippings' folder, inside which I could create subfolders. Actually, that feature is pretty cool :)

But still, why all the love for Bloglines? My initial impressions, based on the first session, were that it sucked. But I think the problem isn't Bloglines, but my expectations for what I should be able to do with an aggregator.

I realised that there are two models for RSS aggregation software. The first, employed by most desktop aggregators, is to treat the software as an interface to a database of entries, where the entry is the atomic unit of RSS -- a single news story or weblog post. They're organised into feeds, but the user can perform operations on individual entries similar to those that might be performed on emails in a mail client -- marking as read, deleting, and perhaps moving them into user-created folders (though I'm not sure many aggregators support this last feature).

The second model is the one employed by most web-based aggregators including Bloglines, and serves the user well if their intent is to check site updates and little else. Little to no control is provided on individual entries -- the atomic unit of this aggregator is not an individual entry, but a feed. In Bloglines, if you click on a feed, it's assumed you've read all the articles it displayed. There's no way to mark only some of them as read, and once you click on a feed, no way even to preserve the entries that were marked as unread a moment ago -- so you'd better read them before you do anything else! And there's definitely no way to delete entries from a feed. Why would you want to do that? It's not your feed!

By now you'll realise that I'm an entry-based aggregator kind of person. I like to read one article at a time, and when I've read it, decide either to delete it from the database or keep it around. I browse feeds in small chunks of time throughout the day while doing other things, not in one big session, so I don't want to be forced to read a whole feed at a time. And I like to keep interesting entries around for later reference (because I couldn't access one of the links the first time, or because I want to refer to it the next time I'm discussing subject X on IRC, or just because I liked the entry a lot... the reason varies). I use SharpReader not only as a viewer, but as a database -- an archive of interesting material in a convenient and organised format that would otherwise use dozens of browser bookmarks. I keep or delete RSS entries like I keep or delete emails -- I don't think the two are all that different in any significant way (which is why, all this talk about server-side aggregation aside, I'm quite excited about Mozilla Thunderbird adding RSS support in its next release).

I think that the power of RSS goes largely to waste if it is just used as an update notification mechanism, something with only transitory utility, useful only at the time of retrieval. I've always seen it as a means to feed a continuing (if sporadic) stream of information into a database, which can then be put to uses beyond simple notification, like search and archival.

I've Googled until sore, and it appears that nobody's made a server-side aggregator of this nature. I'm surprised by this, to the point that I think I must have overlooked one or several such programs. Here's my wishlist for the perfect web-based RSS aggregator. Please, if anything close to this exists, let me know.

• It should be freely available and licensed for personal use, because a personal RSS aggregator is exactly the kind of software people want to run on their own machines and intranets.
• It should be ultra-liberal about feed parsing, because the fact that somebody's feed is broken is of no interest to a reader who probably has no opportunity to fix it.
• And it should understand RSS0, 1, 2, and Atom, obviously.
• It should give fast response times, because a laggy interface is really irritating.
• It should have a friendly and aesthetically pleasing interface, and be easy to use. I feel that Gmail has set the bar for a good web interface for complex functionality, and that the ease of use of Gmail is something to aim for and aspire to.
• It should use no popup windows anywhere, because I hate them :)
• It should be web-based, so it can in theory be accessed from anywhere, and it should be able to operate on the level of individual entries, so a user can start reading where they left off when they finished work an hour earlier.
• It should offer plenty of control over the contents of its database, to appease both the user who is happy using their aggregator merely as a notifier for new content, and the user who also views their aggregator as an information repository. The aggregator should be an interface to a personal database of entries drawn from feeds, and whether and how the user stores or organises this data is up to the user.

Or, Bloglines meets SharpReader meets Gmail. Wow, putting it like that makes it sound like such a straightforward job :)

Would anybody find something like this useful? Or am I in a minority in using an aggregator for anything more than update notification?