I hear Google founders have entered the Forbes list. Makes one wonder, whoa, I'm gonna make a power search engine and get rich.
Oh, if it were that simple.
I have head search engine-like problems in front of me before. I've learnt this one thing... There are two philosophies on making a search engine with a huge amount of data. Either (a) it's slow, and by slow I mean as slow as that one; or (b) it's huge, and by huge I mean as huge as 5 to 10 times larger than the total sum of the original data.
It's that big of a problem.
Which makes me shrug and sigh, but understand that there's a good reason for the nonexistence of an internal LJ search. As well as 95% of the webcomics I read.
...I believe all of you are familiar with the "now where is that post/comment where I/he/she/them said something about that"? The current solution to that is memories, but memories have a fundamental flaw: you must foresee that this post/comment will be needed later. Which is something I'm horrible at. :P
I said all that to tell you why I think search is problematic, yet necessary.
I'd like to see what sort of discussion is over search solutions. I'm currently a free user, and I have three major ideas in mind (business-wise, since that's the spirit of this community).
(1) Free For All
To make searching reliable for all users, the "slow" approach should be necessary, and, more than that, it should be explored. My idea here is that the user enters their query string to search on someone's LJ archive, to whom only visible-to-logged-in-user entries would be returned after one day, or some arbitrary time based on the amount of entries that should be looked to. Similarly, if the logged in user wanted to search to all their friends instead of only one LJ user, a short time span (up to 6 months) should be specified, and the amount of time would be dependent on the amount of friends. I forgot to mention that the link to search results would be e-mailed to the user. :P
(2) Free For Some
This should require an intermediate approach, similar to what seems to be available on the directory. I'm not sure of how that should work specifically because I'm not aware of the performance/space issues as the LJ staff is. Anyway, the idea here is train more monkeys to search the entry bases. :) Only paid users should have the right to search and their search would take a little time. Depending on the time (it would vary), they'd get the "training monkeys" refresh page, or a warning that it's gonna take some time and to watch their mail.
(3) Trading Good
Finally, this is the other extreme. Either the user would pay per search (O_O!) or the user would pay for the right of making searches for a period of time - much like the additional userpics system. I think this is a very risky approach because it'd require the search to be fast, and I have no idea of the acceptance of that. The LJ staff will have a better idea of that, but that would probably still be a wild guess. As a solution, this is what looks least appealing...
So then... These are my 2�. Thoughts? Opinions?