It's not instantly clear from e.g. http://www.lojban.org/publications/etymology/etysample.txt what the algorithm was for the computerized part of the process of gismu construction, although I doubt it was as naive as you seem to be implying. In any event, the humans involved also played with the gloss words they were feeding it to try and find good compromises. {klama} does not in fact take the "l" and "a" from "travel" - it comes from (in Lojban's phonetic spelling) the English "kam" (come), the Hindi "ana" and the Chinese "lai".
At worst, learning Lojban vocabulary is no worse than learning any foreign language, and at best it's somewhat easier (which has been my experience) - I certainly don't see how its phonology "just makes it really difficult for everyone".
At worst, learning Lojban vocabulary is no worse than learning any foreign language, and at best it's somewhat easier (which has been my experience) - I certainly don't see how its phonology "just makes it really difficult for everyone".