Boy, I have fond memories of participating in the Lojban Yahoo group back in 2003 when I was 16. I carried around the Cowan book to my classes and loved reading random sections. I was in awe of how elegantly it was designed, and it inspired me to invent my own languages.
My only complaint about the language was about the reliance on word order instead of prepositions to determine the relationship between nouns. It made the language harder to learn since you have to memorize the parameters for each word, and it always felt arbitrary (what if there is no parameter for the meaning I wish to convey?).
You could always prefix the arguments with their place markers. Then choose to move them around, or leave them in their default positions. Most selbri (predicates) never had many sumti (arguments), and the default first or second were designed (rightly or wrongly) to be the most common. Very like arguments to a function, and then having the option of having named argument.
And if there was no parameter you could add it. There were mechanisms for creating new places with appropriate markers. Sounds like you never got to that part of the language - it was definitely in the more advanced usage, but actually when you came across it in practice your brain tended to do the right thing with it. In that sense the system worked as intended.
Largely speaking it has sunk without trace, but another couple of votes will get it to the "Ask" page, so there's a chance someone who can help me will see it, and I'd appreciate that.
Edit: And almost immediately there's a vote - thank you, I do really appreciate that.
It was fun to learn, and it gave me some ideas that got me going in natural language processing despite that it's not a natural language.
The argument structures were one of the big problems I had with it, too: it seemed like someone coming up with predicates just kept tacking on more arguments that were not at all essential to the predicate, in case someone wanted them. This lacks elegance.
Also, the grammar needs an overhaul. There are hundreds of PEG nodes, some of them are almost the same as others except that things you'd want to say are unexpectedly ungrammatical, and some of them could never be used at all except in an interplanetary spelling bee.
I emphasize that this did not stop me from enjoying it, and much later writing an MIT Mystery Hunt puzzle in it [1]. (warning: Mystery Hunt puzzles are hard)
I went by oskar2379. I did a search and only came across one post from my teenage self, linked below. You can see I was complaining about the predicates back then, too =)
With my programmer hat on, using word order to distinguish the parameters to a selbri feels very Haskell. However, using prepositions feels very Python (i.e. keyword parameters).
I do agree with you about using keyword parameters. The Lojban way to do it would be to add single-syllable cmavo for each position, so that you'd tag parameters depending on what position you wanted them to fill. But this would involve adding at least one syllable to every phrase which involved a selbri, and probably more. It'd be interesting to calculate what that would do to the overall bandwidth.
Hey, guess what! There are single-syllable cmavo for doing exactly this! fa, fe, fi, fo and fu, respectively. Although the docs warn against using them too much because they might be hard to parse.
I remember learning quite a bit of Lojban back in 1989 using a flashcard program on my PC. I even subscribed to Bob LeChevalier's newsletter for a while. I drifted away from it probably because it wasn't solving a problem that I had, cool as it was in theory. It was probably more interesting to construct the language than to use it.
My only complaint about the language was about the reliance on word order instead of prepositions to determine the relationship between nouns. It made the language harder to learn since you have to memorize the parameters for each word, and it always felt arbitrary (what if there is no parameter for the meaning I wish to convey?).