Firstoff, I think, if you're going to create a service that may appeal to a user base of nearly 1(M) Million installed, you really need to be aware of the bandwidth needs you may have.
And let's face it, anything that cleverly runs comparative data analysis on livejournal and returns post-able HTML code is going to get hit relatively hard.
That being said... ljmatch has potential. It's strong off the starting block but has a whole lot of limitations. This may make it a good short sprinter, but really not survive under the endurance race. This would put it in league with the increasingly ignored, "friendster" rather than in the long usable tool category of Marnanel's Friendwatcher
So, what would make ljmatch better.
- First and foremost. Site dependability.
I have seen far too many connection errors and bad database errors to want to push this site
- Some people match at 102%... I think this issue stands for itself.
- If you're going to list religion as a check mechanism...
you HAVE to include a selection for "Other"
and you MUST also make it a series of checkboxes
and you should include a series of radio buttons gauging 'spirituality level' which is far removed from what religion you may have been inflicted with.
- Sexuality as well as gender...
let me tell you exactly how compatible a tenaciously homophobic person is with a "convert them all" person.
- and while we're there...Liberal and Conservative is not merely a test of government... I think that ljmatch does an okay job here... but does fall flat in places. The questions tend to be a bit overly vague...meaning it's the programmer who interprets the relationship between these questions.
But let's be honest here. There really are no good 'matching' systems out there yet. If there were, they'd be rich. The problem is also based on the issue that people's opinions on some of these topics can be very quicksilver in the way they change depending on the answerer's mood, relationship, or intent for the test.
I will give ljmatch points. Looking at my scores and scores between people on my list, I think they did a fair job. I think it could be better because "compatibility" is one of those grossly open-ended terms that means something different for each type of relationship. I may be more than compatible to be a friend and horrendous as your co-worker. You might make a fantastic lover, but our politics would have us at each other's throat. I will admit that the people who are in the top 95% surprise me. In no order: 3 people I don't know well and the other 2 I tend to get along with very well, but tend to believe that I differ from on a personal level at an extreme level.
And I know that there is at least one person on my friends list who will vehemently disagree with this... (waves to "J")
I think that Brad et all, should work with these small sites and incorporate them into LJ. The searching and compatibility mechanisms in LJ have become too unusable. (Most have caps that stop at the first 1000 records, which is now 1/1000 accuracy) I think the mechanisms should be evaluated and worked into the code base to make it more usable on the LJ end rather than adding up a collection of unofficial off site programs that ask for access back into the system.
Just my 50.3 Surinam Guilder's worth....