A few months ago I gave up working on Frassle, the experimental blogging platform that hosts my blog and a few others. Frassle was a fun and rewarding project: it put me in touch with the very vibrant world of bloggers and social software development, it gave me a reason to present at OSCOM in Switzerland, and it even earned a few passionate users. But I ultimately decided to cease working on frassle. A couple of people have recently expressed interest in picking up where I left off, and that's given me reason to reflect on what I've learned and where it might lead next.
The main reason I stopped working on frassle is that I started putting my time into other projects like voo2do (an easy-to-use online task management application, much more popular than frassle ever was) and my day job. Frassle also has some performance problems: the database structure has too many triggers and a seriously slow full-text search system. The frassle studio—which I think is a good idea that still hasn't been done well—turned out to be perform incredibly badly when implemented on top of a relational database, but could perform wonderfully on a custom sort of publish-subscribe DB. These issues left me in a spot where I felt I'd need to do a lot of rebuilding before I could add nifty, user-visible features.
Yet overall, the core problem with frassle is that it tries to do too many things. It makes conceptual sense to integrate blog publishing, aggregators, republishing, and semantic correlation. But it doesn't make practical sense. In trying to tackle all of these issues at once, I was never able to be the best at any of them. That weakness almost completely prevented users who tried frassle from ever coming back, because when a user is trying a web app, they're not looking for long-term conceptual potential. They're looking for tools that solve a specific problem or make life noticeably better in a matter of minutes. If an app doesn't walk a user through a pleasant experience in the first 5 minutes, that user will not come back. And if you can complete the experience in less than 5 minutes, you will roughly double user retention for each minute saved. I didn't fully appreciate this adage until voo2do, which took a lot less novel thinking and implementation time than frassle, got more users in a week than frassle had in two years.
So what would I do differently if I could do it again? Probably one of the following three mutually exclusive goals:
1. Focus on the inter-blog category mapping feature set only. Create a form where you can paste a blog URL and get content from other blogs that is relevant to the given blog, or specifically relevant to categories within that blog. Use the algorithms from
http://frassle.net/help/welcome and a huge database of RSS feeds with a custom aggregator/analysis layer; a relational database would even work OK.
2. Build a better Frassle studio. Everyone produces RSS now; sites like Pageflakes and Netvibes and Google Customized Homepage are a dime a dozen; but none of these sites let you weave content from multiple sources into a new community website. At some point the people making these tools are going to realize that they're awfully close to fulfilling a non-solipsistic need, and Google Customized Community Website will be born (or bought). Just mix together the tiles-of-content model with the ability to create custom message boards, and you'll have a community website contruction system that offers more power than the average blog, but isn't much harder to user.
3. Social intranet in a box. Now let me partially retract my earlier advice. The one market that *would* be well served by an integrated suite of blogging/sharing tools, even if they weren't all best-in-class apps, is the intranet market. There are tons of LAN-connected collaborative organizations whose members don't want to communicate on the public internet. If you can give these people a bootable CD that they can pop into a spare PC, turn on, and instantly start using as a collaborative intranet server, they would line up around the block to worship you. Nobody gives much thought to usability in intranet applications because the sales cycle is slow, expensive, and managed by IT departments who never use the software and thus don't care if it's excruciating. But if you made bootable CDs that tech-savvy non-IT folks could pop into a spare PC, you could seed a revolution in intranet apps that don't suck.
Sometimes I feel like frassle was my big chance. It had some of the ideas of tagging and social bookmarking even before del.icio.us, and if I'd thought to focus on those parts, maybe I'd have a top 500 site, a yacht from Yahoo!, and 15 million blog groupies. But if frassle fell short because my practical skill didn't match my creative vision, I can only hope that I've learned enough to be get things done without sacrificing much of that vision. I'll just have to work and see.